ACCUWIND - Methods for Classification of Cup Anemometers (RIS

Transcription

ACCUWIND - Methods for Classification of Cup Anemometers (RIS
Risø-R-1555(EN)
ACCUWIND - Methods for Classification
of Cup Anemometers
J.-Å. Dahlberg, T.F. Pedersen,
Peter Busche
Real
cup-anemometer
Calibrations
-friction
-torque curves
-inertia
-angular char.
Cup-anemometer
model
Fitting to
Environmental
cup-anem.
operational
model
conditions
Classification
index
Calculation
of responses
Class
Risø National Laboratory
Roskilde
Denmark
May 2006
Abstract Errors associated with the measurement of wind speed are the major
sources of uncertainties in power performance testing of wind turbines. Field
comparisons of well-calibrated anemometers show significant and not acceptable difference. The European CLASSCUP project posed the objectives to
quantify the errors associated with the use of cup anemometers, and to develop
a classification system for quantification of systematic errors of cup anemometers. This classification system has now been implemented in the IEC 6140012-1 standard on power performance measurements in annex I and J. The classification of cup anemometers requires general external climatic operational
ranges to be applied for the analysis of systematic errors. A Class A category
classification is connected to reasonably flat sites, and another Class B category
is connected to complex terrain, General classification indices are the result of
assessment of systematic deviations. The present report focuses on methods that
can be applied for assessment of such systematic deviations. A new alternative
method for torque coefficient measurements at inclined flow have been developed, which have then been applied and compared to the existing methods developed in the CLASSCUP project and earlier. A number of approaches including the use of two cup anemometer models, two methods of torque coefficient
measurement, two angular response measurements, and inclusion and exclusion
of influence of friction have been implemented in the classification process in
order to assess the robustness of methods. The results of the analysis are presented as classification indices, which are compared and discussed.
ISBN 87-550-3514-0
Risø-R-1555(EN)
2
Contents
Preface 4
1.
Introduction 6
2.
Classification Procedures 6
2.1 Classification Principles 6
2.2 Classification index 8
2.3 Defining Measured Wind Speed 9
3.
Laboratory and Wind Tunnel Tests – Calibrations 10
3.1 Equipment 10
3.1.1Wind Tunnel LT5-II 10
3.1.2The tilt device 11
3.1.3The gust generator 11
3.1.4Light propellers to measure instantaneous wind speed 12
3.2 Conversion of pulse trains to wind speed and angular speed 12
3.3 Calibration of the tooth wheel signature 13
3.4 Normal Calibration of the Anemometer and the four propeller
anemometers 14
3.5 Tilt Response Measurements 16
3.6 Static Torque Measurements at Horizontal flow and Forced OffEquilibrium Speed Ratio 18
3.6.1Torque measurements on the classical Thies cup anemometer. 18
3.6.2Torque measurements on RISØ P2546 cup anemometer. 19
3.7 Dynamic Torque Coefficient Measurements at Horizontal Flow 22
3.7.1Gust tests with the classical Thies cup anemometer 22
3.7.2Derivation of torque characteristics 25
3.7.3Comparison of Cq derived from gust tests and torque measurements 26
3.7.4Discussion 27
3.8 Combined tilt and ramp-gust tests 28
3.9 Comparison of dynamic torque measurements with angular response
measurements 29
3.10 Gust Run Measurements at Horizontal Flow 30
3.11 Step Response Measurements 32
3.12 Bearing Friction Measurements 34
3.13 Rotor Inertia Measurements 39
4.
Time Domain Modelling of Cup-Anemometers 40
4.1
4.2
4.3
4.4
Introduction 40
Generalization of the rotor torque 40
The Tilt-Response & Torque-Coefficient Model (TRTC) 42
The Inclined-Flow-Torque-Coefficient Model (IFTC) 45
5.
External Conditions for Cup Anemometer Classification 46
5.1 External Operational Conditions 46
5.2 Wind speed range 47
5.3 Turbulence ranges 47
5.3.1Turbulence intensity levels 47
5.3.2Turbulence spectra 48
5.4 Air temperature ranges 50
Risø-R-1555(EN)
3
5.5 Air density range 50
5.6 Flow inclination angle ranges 50
5.7 Artificial 3D Wind Generators 51
6.
Application of Classification Procedures 52
6.1 Classification of cup anemometers 52
6.2 Robustness of applied procedures 52
6.3 Calibration Robustness 52
6.3.1Differences in normal calibrations 52
6.3.2Differences in tilt response measurements 53
6.3.3Differences due to inertia 55
6.4 Cup Anemometer Model Robustness 56
6.4.1Simulation of normal calibration 56
6.4.2Comparison of measured and calculated step response 57
6.4.3Comparison of measured and calculated response from gust runs 57
6.5 Robustness of Classification 58
6.6 Free Field Comparison Tests 61
6.6.1Free Field Tests Compared to Simulations 61
6.6.2Influence of Free Field Horizontal Shear Flow 62
4
7.
Conclusions 63
8.
References 66
Risø-R-1555(EN)
Preface
This report presents results of a European research project ACCUWIND on
improvement of evaluation and classification methods of cup and sonic anemometry. The report presents results of task 1 on cup anemometry. The six ACCUWIND partners are:
o Risø National Laboratory, RISØ, Denmark
o Deutsches Windenergie Institut Gmbh, DEWI, Germany
o Swedish Defence Research Agency, FOI, Sweden
o Centre for Revewable Energy Sources, CRES, Greece
o Energy Research Centre of the Netherlands, ECN, the Netherlands
o Universidad Politechnica de Madrid, UPM, Spain
The work was made under contract with the European Commission, project
number NNE5-2001-00831. Additional support to this part of the project was
given by Risø (Risø National Laboratory, DK) and FOI (Swedish Defence Research Agency, SE)
Risø-R-1555(EN)
5
1. Introduction
The cup anemometer is the standard instrument used for mean wind speed
measurement in wind energy. It is being applied in very high numbers around to
world on masts for wind energy assessments. It is applied exclusively for accredited power performance measurements for certification and verification
purposes, and for purposes of optimisation in research and development. In the
area of verification manufacturers and wind farm owners sign contracts that
specify clearly when a power performance evaluation is acceptable or not. The
use of one type of cup anemometer might give acceptance of the performance
while another one does not. The EU project SITEPARIDEN [1,2,3,4] showed
measured differences between various types of cup anemometers up to 4%,
which is significant when the limits of power performance acceptance is 95% of
guaranteed. The EU project CLASSCUP [5,6] proved that the reason for these
differences were angular response, dynamic effects and bearing friction characteristics, which differs on various types of cup anemometers.
This ACCUWIND report on cup anemometry deals with establishment of robustness in classification by focus on methods and procedures for assessment of
characteristics of cup anemometers. The methods and procedures provide a platform for use in the IEC61400-12-1 standard on power performance measurements [7], as well as for development of improved instruments.
The improvement of accuracy of wind speed measurements is not an easy task.
Uncertainties in each aspect of measurement, from detailed design of sensor,
calibration, application on tall masts, to operational characteristics for wide
ranges of climatic conditions all ads up to high uncertainties.
The problems addressed in this report are subject to a number of innovative
methods. An advanced wind tunnel procedure with dynamic inflow is applied to
the sensors to measure indirectly the torque characteristics without physical interference with the cup anemometer rotor, robust rotor torque measurements on
cup anemometers are developed, and dynamic vertical inflow characteristics of
cup anemometers are quantified.
The present report primarily focus on assessment methods for robust classification of cup anemometers, while a second report [8] presents results of the methods being applied on five different types of cup anemometers.
2. Classification Procedures
2.1 Classification Principles
Evaluation of cup anemometers must be related to the applications, under which
they are applied in the field. An evaluation may be very detailed, and in practice
not easy to apply, but the evaluation can be simplified and made easier to the
6
Risø-R-1555(EN)
user when a classification procedure is applied to the evaluation. A classification procedure has the purpose of categorizing the application in terms of accuracy levels for specified external operational conditions. The task for a user is
then reduced to check the external operational conditions, decide on the accuracy level with the class number, and then select the cup anemometer that corresponds to the required class.
Selection of power converters is very easy due to such a classification system
for electric monitoring equipment. The classification of electric power converters is based on requirements of accuracy of the instruments being lower than a
certain level (classification level) for well-defined operational ranges. In specifying the power measurement equipment for a power performance test of a wind
turbine, the test engineer should look at whether the claimed uncertainty classification of the measurement equipment, particularly the current and voltage
transformers and the power transducer, can be retained. Given the nature of the
measurands, i.e. the expected currents and voltages, the effect on the related
variables should be critically analysed.
The International Electrotechnical Commission has published class indices for
power monitoring and related equipment. Power transducers are covered by IEC
60688: 1992 [9], current transformers IEC 60044-1:1996 [10], and voltage
transformers by IEC 60044-2:1997 [11]. In the case of a power transducer a
classification of 0.1, 0.2, 0.5 or 1 indicates that the limits of intrinsic error will
be within ±0.1%, ±0.2%, ±0.5% or ±1.0% where the ‘fiduciary’ value is the
span. In the case of current and voltage transformers, similar classifications apply.
In order to provide a classification system of cup anemometers similar in
principle to the classification system of electric measuring devices, the external
operational conditions shall be well-defined and operational limits for the
classification shall be set up. Testing of cup anemometers by applying turbulent
winds with variations in all relevant external operational conditions is very
difficult. An easier task is to determine the influence of various parameters of
external operational conditions under idealized conditions. In the laboratory,
fundamental physical characteristics can be determined. The influence of
temperature on friction can be found in climate chambers. The rotor inertia can be
estimated by an oscillatory vibration method. Aerodynamic characteristics can be
found by wind tunnel investigations under quasi-static or dynamic conditions. All
tests, obviously, must assume that these conditions are descriptive for the
characteristics of the cup-anemometers under natural field conditions. Under all
circumstances, the tests made under well-controlled conditions are fundamental in
prescribing the physical behaviour of the cup anemometers. The behavioural
characteristics in the laboratory or in the wind tunnel do not change when the cup
anemometer is put into the field with turbulent wind. The only problem is,
whether the turbulent wind is influencing the cup anemometer in a more complex
way than can be applied under laboratory conditions.
The three most important characteristics of cup anemometers are the dynamic
effects, the angular characteristics and friction in bearings. Time domain cup
anemometer models that are able to handle at least these characteristics
properly, can be used in a classification procedure, for instance as described in
Figure 2-1 [12]. The basic elements are robust calibration procedures, and
appropriate time domain cup anemometer models. An important task is then to
fit calibration data to the models so that the models at least are able to simulate
the calibration tests, as well as other wind tunnel tests, like step responses and
Risø-R-1555(EN)
7
dynamic measurements. When the models are tuned with the cup anemometer
data, they can adequately be applied to calculate responses of realistic
environmental operational conditions.
Real
cup-anemometer
Calibrations
-friction
-torque curves
-inertia
-angular char.
Cup-anemometer
model
Fitting to
Environmental
cup-anem.
operational
model
conditions
Classification
index
Calculation
of responses
Class
Figure 2-1 Elements of a of cup-anemometer classification from laboratory
tests through time domain modelling of the cup anemometer, and calculation of
responses from a modelled three dimensional wind
2.2 Classification index
The class index levels for specified external operational conditions of a classification procedure should be expressed in a way such that it is easy to derive an
uncertainty estimate from a classification index number. The class index system
for power converters [9] is based on systematic variations of the output signal
due to variations in influence quantities, and the class index numbers are stated
as integer numbers ranging over decades of levels of power. For cup anemometers the relevant class indices would not be ranging over decades, and the interesting class index numbers are within a range from zero to two or three. Therefore it is relevant to keep the class index numbers on a proportional real number
scale using decimals, as proposed in the IEC standard [7]. For cup anemometers
the class index numbers are not expressed as a percentage error or as an absolute error value alone because they normally have higher relative uncertainties
at lower wind speeds than at high wind speeds. If percentage errors were selected for the class index number the errors at the lowest wind speeds would
determine the index number. If, on the other hand, absolute errors were selected
8
Risø-R-1555(EN)
for the class index number it would be the highest winds that would dominate.
In these cases it would the systematic errors in the lower wind speed range or
high wind speed range, respectively, that would dominate the classification. The
compromise is an index range, as stated in the IEC standard [7], which combines the absolute and the relative deviations to the class index k :
k = 100 ⋅ max
εi
U i / 2 + 5m / s
Where
Ui
is wind speed in bin i
εi
is the maximum systematic deviation within wind speed bin i
The systematic deviations ε i can be derived from variations of the influence
quantities: turbulence intensity, turbulence structure, air temperature, air density
and average flow inclination angle. The deviations can then be plotted in a
graph with the class limits as shown in Figure 2-2.
Permissible Deviations [m/s]
Permissible Limits of Error for Cup Anemometer Class Indices
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
-0,1
-0,2
-0,3
-0,4
-0,5
-0,6
-0,7
-0,8
Class 0.5
Class 1
Class 2
Class 3
Class 5
0
2
4
6
8
10
12
14
16
18
20
Wind Speed [m/s]
Figure 2-2, IEC61400-12-1 classification class index examples
2.3 Defining Measured Wind Speed
The measured wind speed of a cup anemometer is an averaged one-dimensional
quantity. The interpretation of this quantity, though, is important in order to
make a consistent classification.
Considering a time dependent 3D wind speed vector with a longitudinal
component u, a transversal component v, and a vertical component w as input to
the cup-anemometer.
r
U = ( u , v , w)
The measured wind speed of a cup anemometer, is in the IEC standard [7] defined
as the “horizontal” wind speed. This definition includes only measurement of the
horizontal wind speed components (length of the wind speed vector, excluding
the vertical component):
Risø-R-1555(EN)
9
U hor =
1
T
∫
u 2 + v 2 dt
t
Or expressed as an averaged digitized measurement:
U hor
1
=
N
N
∑
i =1
ui2 + vi2
If a cup anemometer has a cosine angular response, the vertical component w is
filtered away, in which case it automatically covers the “horizontal” wind speed
definition.
A vector wind speed is defined as the scalar of the “vector” of wind speed: This
definition includes all wind speed components (the scalar of the wind speed
vector):
U vec =
1
T
∫
u 2 + v 2 + w 2 dt
t
Or expressed as an averaged digitized measurement:
U vec =
1
N
N
∑
i =1
ui2 + vi2 + wi2
An analysis of the influence of the definition of the measured wind speed on
power curve measurements was made in [13].
3. Laboratory and Wind Tunnel
Tests – Calibrations
3.1 Equipment
3.1.1
Wind Tunnel LT5-II
All tests described in this report, except for DEWI tilt tests, referenced in chapter 6.3.2, were carried out in FOI wind tunnel LT5, which fulfils the recommendations of IEA for calibration of most used cup anemometers. The wind tunnel
is shown in Figure 3-1. Tunnel data are presented in Table 3-1.
10
Risø-R-1555(EN)
Figure 3-1 The large test section of LT5. The air is sucked from right to left
and is blown out through a window in the back.
Table 3-1 Wind tunnel data of FOI LT5-II (large test section)
Test section height:
Test section width:
Test section length:
Test section cross section:
Speed range:
Turbulence:
3.1.2
0.675m
0.900m
2.5m
0.61m2
~0-16 m/s
< 1%
The tilt device
To measure the response from a cup anemometer that is exposed to non horizontal flow the pole that holds the anemometer can be tilted ±50º. The tilt tests
are usually carried out such that the anemometer is slowly tilted back and force
during 20 minutes. During this sequence the wind speed, tilt angle and rotational speed of the anemometer is sampled continuously. The data is processed
by binning the data by tilt angle bins, and averaging the relative anemometer
speed to horizontal speed.
3.1.3
The gust generator
Wind gusts can be simulated by varying the air speed of the tunnel sinusoidally.
This is accomplished by means of two plates located in the outlet of the fan according to Figure 3-2.
The plates are driven by means of a three phase electric motor which is fed from
a frequency converter such that the gust frequencies can be varied from 0 to 5
Hz. The plates can be driven in different ways; by keeping one plate fixed the
turbulence intensity can be varied from 10 to 22 %; by rotating both plates a
turbulence intensity of 32% can be reached.
Risø-R-1555(EN)
11
Figure 3-2 Details of the gust generator with the plates closed or opened. The
plates can be driven by a chain transmission.
3.1.4
Light propellers to measure instantaneous wind speed
Special light propellers have been developed to measure rapid changes of the
wind speed, see Figure 3-3. They are stream shaped from thin balsa wood and
have each a weight of about 0.07 gram. The 1 mm shaft is supported by small
miniature bearings. The rotational speed of each propeller is measured by
equipment that consists of a laser beam and a photo sensor connected to an electronic device. The beams are directed from the outside of the tunnel through the
plane of rotation of the test objects and points at the photo sensors. The rotating
object (anemometer or propeller) cuts the beam and TTL pulses are creates as
output.
Figure 3-3 Close up view of prop 3 (left) and comparative tests to select the
“best” propellers
3.2 Conversion of pulse trains to wind speed and
angular speed
The conversion from pulse train frequency for the anemometer and the propellers to corresponding wind speeds are based on the calibration transfer function
determined by an ordinary calibration without gusts. The instantaneous wind
speed, during gust events, can be calculated from the time difference between
two consecutive pulses with a time resolution given by the frequency of the
12
Risø-R-1555(EN)
pulse trains. However this method gives, due to inevitable deviations in the design of the tooth wheel, rise to significant and unacceptable scatter.
Theoretical investigations, based on an ideal anemometer, showed that it was
necessary to have an, as accurate as possible, information of the rotational speed
of the anemometer. This implies that it is necessary to have detailed information
of the individual tooth position along the revolution. This information needs to
be determined by calibration of each individual anemometer.
3.3 Calibration of the tooth wheel signature
The calibration can be carried out in two ways. Either with a fly wheel in still
air or with the anemometer exposed to constant wind speed of different magnitudes as an ordinary calibration. For both methods the time between pulses are
averaged and normalized.
The first method gives an absolute measure of the gap distribution (signature) of
the tooth wheel and the second method gives a combined result which also includes the variations in rotational speed caused by different aerodynamic forces
from the position of the cups. The latter method will give a signature that is dependent on the direction of the wind flow in relation to the fixed part, the house.
In the following Figure 3-4, Figure 3-5 and Figure 3-6, graphs are presented to
demonstrate the results from calibration of the tooth wheel of the classical Thies
anemometer.
Thies (disc) ccw
1,06
1,04
disc_134634
disc_134652
1,02
relative gap
1,00
0,98
0,96
0,94
0,92
0,90
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
tooth #
Figure 3-4 Absolute measure of the tooth wheel signature by means of disc
rotation in still air. Two 10-second measurements with 100 kHz clock pulse frequency for the timing of the pulses. Note that the difference between the smallest and widest gap is about 10%.
Risø-R-1555(EN)
13
Profile from runs with Thies in 8 m/s
1,06
140207_8
1,04
140425_8
1,02
Relative gap
1,00
0,98
0,96
0,94
0,92
0,90
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
tooth #
Figure 3-5 Determination of the combined tooth wheel signature derived from
two 100-second measurements in 8 m/s wind speed.
Profile from runs with Thies in 8 m/s, corrected from disc runs
1,06
1,04
140207_8_disc corrected
140425_8_disc corrected
1,02
relative gap
1,00
0,98
0,96
0,94
0,92
0,90
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
tooth #
Figure 3-6 Determination of the air component in the tooth wheel signature as
a result from a combination of the signatures in the figures above. The three per
rev variation appears now very clearly.
3.4 Normal Calibration of the Anemometer and
the four propeller anemometers
All tests are preceded by a normal calibration of the anemometer together with
the four light propeller anemometers, see Figure 3-7. The propeller anemometers are used to measure the instantaneous wind speed during gust tests.
14
Risø-R-1555(EN)
Figure 3-7 A typical setup of the anemometer and the four propeller anemometers in the LT5 wind tunnel. The wind is blowing from right to left. The
upper propellers are labelled p2 and p3 from left to right and the lower propellers are labelled p4 and p5 from left to right.
The calibration procedure follows in principal an ordinary calibration. Each data
point is acquired during 125 seconds. The measurement sequence starts and
ends with a zero reading with the tunnel shut down. The first point is 5 m/s then
7, 9 etc. up to 15 m/s. On the way down the even wind speeds are measured.
The analogue volt values from the WT pressure transducer and tilt angle sensor
are sampled at 100 Hz. The pulses from the anemometer and the propellers are
clocked with a 100 kHz timer.
The evaluation procedure involves; counting of pulses, conversion of pressure
signals to wind speeds and straight line fitting. A typical result is shown in
Table 3-2 and Figure 3-8.
Table 3-2 Typical calibration values of the four propeller anemometers
Object
Anem
P2
P3
P4
P5
Risø-R-1555(EN)
Slope Intercept
R2
St.err.
0,0461
0,151
0,99996 0,035
0,0940 -0,053 0,99999 0,017
0,1036
0,195
0,99998 0,025
0,1128 -0,017 1,00000 0,011
0,1214
0,052
0,99994 0,040
15
18
16
CLI : Slope 0,0461 : Offset 0,151 : Correl 0,99996 : St.err 0,035
Tunnel Speed [m/s]
14
12
10
CLI :
fit
8
6
4
2
0
0
50
100
150
200
250
300
350
400
Pulse frequency [Hz]
Figure 3-8 A typical result from calibration of a Climatronics anemometer
over the 4-15 m/s range.
3.5 Tilt Response Measurements
During the tilt tests the anemometer is slowly (approximately 2º/sec) tilted back
and forth during 1050 seconds. The tilt angle and tunnel speed are continuously
measured with 20Hz and the pulses from anemometer and the four propellers
are sampled with a 100 kHz clock timer. The evaluation involves conversion
from pulse frequencies to wind speed by using the transfer functions from the
calibration. The data are sorted into 2º tilt bins and averaged. The tilt tests are
performed at wind speeds of 5, 8 and 11 m/s. Figure 3-9 shows a typical test
with the cup anemometer in the extreme tilt positions. Figure 3-10 shows typical time traces of tilting angle and sensor output. Figure 3-11 shows binned results of the test, and Figure 3-12 the final angular responses of the tilt test.
Figure 3-9 The anemometer in the -47º and +47º turning-points. Wind is blowing from right to left.
16
Risø-R-1555(EN)
11,0
40
10,5
30
10,0
20
9,5
10
9,0
0
8,5
-10
8,0
-20
7,5
tilt_01_141553
CLI_01_141553_8m/s
-30
-40
7,0
6,5
-50
0
200
400
600
800
Anemometer response
Tilt angle
50
6,0
1200
1000
Time [s]
Figure 3-10 Time traces from the test with Climatronics sensor at 8 m/s wind
speed. The blue curve shows the tilt angle and the red curve the anemometer
response. No tilt motion is performed during the first 60 seconds to ensure that
sufficient number of values fall into the zero tilt angle bin (±1º) which, during
the evaluation, is used for normalisation.
Tilt tests with Climatronics
14
Indicated wind speed
12
10
8
6
CLI_8
CLI_8R
CLI_11
CLI_5
4
2
0
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt angle
Figure 3-11 Binned results from tilt tests with Climatronics sensor at four
wind speeds.
Risø-R-1555(EN)
17
Tilt tests with Climatronics, 050901
1,15
CLI_11_norm
CLI_8_norm
CLI_8R_norm
CLI_5_norm
cos
Relative Speed
1,10
1,05
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt angle (neg from below)
Figure 3-12 Normalised binned results from tilt tests with Climatronics sensor
at all wind speeds. The values are normalised with the value in the zero bin. The
anemometer significantly “over reads” the wind speeds at high negative (wind
from below) inflow angles.
3.6 Static Torque Measurements at Horizontal
flow and Forced Off-Equilibrium Speed Ratio
This chapter refers to a torque measurement method developed and performed
prior to the ACCUWIND project [5]. The measurements refer to two cup anemometers with quite different dynamic characteristics, i.e. different torque characteristics [5,14]. The torque measurements in this chapter have not been normalized, but they can easily be normalized by the use of the normalization
method described in Chapter 4.2.
3.6.1
Torque measurements on the classical Thies cup anemometer.
The torque characteristics of the classical Thies anemometer was determined by
measurements carried out in the CLASSCUP project in February 2001 [5]. For
those tests an external motor was driving the anemometer through a thin shaft
through the wind tunnel wall, and attached to the rotor, see Figure 3-13 and
Figure 3-14. The motor was attached to a linkage system such that the reaction
torque could be measured.
The test was carried out such that the anemometer was exposed to constant
wind speed of 8 m/s and the motor was used to drive or be driven by the anemometer and thereby force the anemometer to operate of-equilibrium.
For each setting of rotational speed all necessary parameters were acquired with
60 Hz sampling frequency during 30 seconds. The rotational speed was measured by counting pulses from the anemometer.
Figure 3-15 presents the derived torque to angular-speed relationship.
18
Risø-R-1555(EN)
Figure 3-13 The pictures show the measurement setup for the measurement of
torque from the Thies classical in LT5 2001. In the right picture the motor
(black) and the thin shaft from the motor shaft to the hub of the anemometer
rotor can be seen.
Figure 3-14 In the left picture the balance to measure the reaction torque can
be seen.
Measured Torque for Thies Classical
30
20
Torque [mNm]
10
0
-10
-20
-30
-40
0
10
20
30
40
50
Angular Speed [Radianer/sec]
Figure 3-15 The derived torque to angular-speed relationship of Thies classical cup anemometer.
3.6.2
Torque measurements on RISØ P2546 cup anemometer.
The measurement of static torque characteristics on a RISØ P2546 cup anemometer was made at FOI in Sweden with the use of a special balance, accord-
Risø-R-1555(EN)
19
ing to [14]. A thin rod was attached to the top of the cup anemometer (seen in
Figure 3-16). The rod was passed through the top of the tunnel roof (small hole
in white paper) and attached to a motor (dark rotational symmetric body, see
Figure 3-17). The motor was attached to a shaft that was mounted in two bearings so that the motor could rotate. The shaft was held in place by a balance,
mounted on a clamp on the shaft. The balance was calibrated after the measurements.
Figure 3-16 Test set-up for rotor torque measurements with thin rod mounted
on top of cup anemometer and going through hole in ceiling to motor and strain
gauge bridge measurement system mounted above transparent wind tunnel ceiling
20
Risø-R-1555(EN)
Figure 3-17 Torque measurement system with motor in torpedo like housing,
bearings on spring loaded mounting rod in grey housing and spring balanced
strain gauge bridge attached to the mounting rod to the left of the light grey
attachment fixture
Torque measurements were carried out at 8 m/s at various rotational speeds,
concentrated around equilibrium tip speed ratio. For each point a 30sec measurement was made. The results are shown in the following Figure 3-18.
Measured Torque of RISØ P2546 at FOI-LT5 at 8m/s
10
8
6
4
Torque mNm
2
0
-2
-4
-6
-8
-10
-12
-14
-16
15
20
25
30
35
40
45
50
55
60
65
70
Angular speed rad/s
Figure 3-18 Torque measurements at FOI at 8m/s on RISØ P2546 cup anemometer [14]
Risø-R-1555(EN)
21
3.7 Dynamic Torque Coefficient Measurements at
Horizontal Flow
The dynamic torque coefficient measurement method is a new method involving exposure of the anemometers to wind gusts in a wind tunnel together with
accurate measurements of the instantaneous wind speed and of the rotational
speed of the cup-anemometer. This method requires no attachment of a rod to
the rotor as for the static torque coefficient measurement, but it requires a much
higher resolution of the rotational speed. The torque-characteristics (normalised
torque coefficient versus speed ratio, see Chapter 4.2) are derived indirectly
from these measured time traces.
The speed sequences from the anemometer and the propellers are asynchronous
in time and there is no synchronisation between the anemometer and the propellers. This is solved by linear interpolation to a common fixed time scale which
is updated at a high frequency.
Once the instantaneous wind speed in the vicinity of the cup anemometer and
the instantaneous angular speed of the anemometer are known, the normalised
torque characteristics can indirectly be determined from the time series according to the following method.
Assuming friction to be negligible, the torque and the speed ratio are:
I
dω
1
= Q = ρU 2 ARCQ
dt
2
ωR
λ=
U − Ut
Where:
I
ω
t
Q
ρ
U
Ut
A
R
λ
rotor inertia
angular speed
time
torque
air density
wind speed
threshold wind speed
cup area
radius to cup centre
speed ratio
From the general torque equation, a stepwise torque coefficient can be derived
for small time steps, assuming constant torque over one time step.
CQ ≈
3.7.1
Δω 2 I
Δt ρ ARU 2
Gust tests with the classical Thies cup anemometer
Tests were carried out with the classical Thies anemometer in May 2004. The
tests involved ordinary calibration in the wind speed range 5-15 m/s, gust tests
with various gust frequencies and one so called ramp test where the gust frequency varied linearly up and down during several minutes. Test setup is shown
in Figure 3-19.
22
Risø-R-1555(EN)
Figure 3-19 The set-up with the classical Thies cup anemometer in the centre
surrounded by four propeller anemometers.
Wind sequences from air-signature corrected gust tests
The classical Thies anemometer has a tooth wheel which gives 44 pulses per
revolution. The four-bladed propellers give 4 pulses per revolution. The anemometer gives typically 160 Hz and the propellers 80Hz pulse frequency at 8
m/s wind speed.
Once the tooth wheel signature derived from exposure in a constant air stream
are known, all pulse measurements can be corrected such that unintentional
variations in the pulse frequency caused by imperfections or cup positions can
be eliminated.
In the following, results from the evaluation of the gust tests are presented.
Unless otherwise stated, the following assumptions hold:
o The pulse frequencies from the anemometer were corrected with the
tooth wheel signature averaged from all tests.
o The wind speed (angular speed) of the anemometer was based on the
time difference between four consecutive pulses.
o The “true” wind speed, i.e. the wind tunnel speed was determined as the
average of the four propeller anemometers. Their (angular speed) was
based on the time difference between four consecutive pulses, i.e. complete revolutions.
Figure 3-20 shows a typical time sequence from gust tests at a constant frequency. Figure 3-21 shows a typical time sequence from a ramp gust test, and
Figure 3-22 shows a smaller tooth-wheel signature corrected time sequence of
the ramp gust test.
Risø-R-1555(EN)
23
11
10
Windspeed[m/s]
9
8
anem_173725_0.034Hz
props_173725_0.034Hz
anem_175642_1.347Hz
props_175642_1.347Hz
7
6
5
0
5
10
15
Time [seconds]
20
25
30
100
120
Figure 3-20 Typical time sequences from the gust tests.
11
Windspeed [m/s]
10
9
8
7
6
5
0
20
40
60
Time [seconds]
80
Figure 3-21 Two minutes of time sequences from the ramp gust test.
10
anem_181812_corrected
anem_181812_not corrected
Wind Speed [m/s]
9,5
9
8,5
8
7,5
100
101
102
Time [seconds]
Figure 3-22 Graph of two seconds of time sequences from the previous graph
with un-corrected and tooth-wheel signature corrected signals.
24
Risø-R-1555(EN)
3.7.2
Derivation of torque characteristics
The torque characteristics can now be derived from the time series according to
the formula described above:
CQ ≈
Δω 2 I
Δt ρ ARU 2
The figures from Figure 3-23 to Figure 3-26 show typical results from the
evaluation.
4
3
2
1
0
-1
-2
-3
cq_181812_ramp_not corr
cq_181812_ramp_corr
-4
-5
-6
0,25
0,3
0,35
0,4
0,45
0,5
0,55
0,6
TSR
Figure 3-23 Cq derived from the corrected and the un-corrected signals. TSR
is equal to speed ratio λ
1,0
cq_175642_1.347Hz
0,5
cq_175402_1.000Hz
cq_175125_0.666Hz
0,0
cq_174836_0.334Hz
Cq
cq_174601_0.202Hz
-0,5
cq_174253_0.135Hz
cq_174007_0.068Hz
cq_173725_0.034Hz
-1,0
-1,5
-2,0
0,25
0,30
0,35
0,40
0,45
0,50
0,55
0,60
0,65
0,70
TSR
Figure 3-24 The graph presents a scatter plot of Cq derived from major parts
of the gust runs. The gust frequencies are given in the legends.
Risø-R-1555(EN)
25
0,6
cq_173725_0.034Hz
0,4
cq_174007_0.068Hz
cq_174253_0.135Hz
0,2
cq_174601_0.202Hz
cq_174836_0.334Hz
0,0
cq_175125_0.666Hz
-0,2
Cq
cq_175402_1.000Hz
cq_175642_1.347Hz
-0,4
-0,6
-0,8
-1,0
-1,2
0,25
0,30
0,35
0,40
0,45
0,50
0,55
0,60
TSR
Figure 3-25 The graph presents Cq as a result of binning and averaging of the
scatter from the previous graph.
0,6
0,4
0,2
all together, narrow bins
all together, wide bins
cq_181812_ramp
0,0
Cq
-0,2
-0,4
-0,6
-0,8
-1,0
-1,2
0,25
0,30
0,35
0,40
0,45
0,50
0,55
0,60
TSR
Figure 3-26 The graph presents Cq as results of binning and averaging of all
data with wide and narrow bins and as the result from binning of the ramp test.
3.7.3
Comparison of Cq derived from gust tests and torque measurements
Torque curves derived by direct static torque measurements or by indirect dynamic measurements by gust runs can now be compared. A smaller correction
due to variability of the wind tunnel was made.
Figure 3-27 shows a comparison of torque measurements on a Risø P2546 cup
anemometer with the two different methods. Figure 3-28 shows a similar comparison for the classical Thies cup anemometer.
26
Risø-R-1555(EN)
Torque measurements on Risoe P2546a cup anemometer @ 8 m/s
0,8
measured with balance
derived from gust runs (2,4,1)
0,6
0,4
Cq(tsr)
0,2
0,0
-0,2
-0,4
-0,6
FOI
-0,8
-1,0
0,10
0,15
0,20
0,25
0,30
0,35
Tip Speed Ratio
0,40
0,45
0,50
Figure 3-27 The graph shows a comparison of Cq for the Risø P2546 cup
anemometer measured with the static and the dynamic torque measurement
methods. Speed ratio from dynamic measurement has been offset by -0.006 and
Cq scaled up by 7% (rotation around Cq=0)
Torque measurements on Classical Thies anemometer @ 8 m/s
0,6
0,4
measured with balance
0,2
derived from gust runs (4,4,1)
Cq(tsr)
0,0
-0,2
-0,4
-0,6
-0,8
-1,0
0,20
0,25
0,30
0,35
0,40
0,45
0,50
0,55
0,60
Tip Speed Ratio
Figure 3-28 The graph shows a comparison of Cq for the classical Thies cup
anemometer. Speed ratio from dynamic measurement has been offset by +0.007
and Cq scaled by 6% (rotation around Cq=0)
3.7.4
Discussion
The results presented show a good agreement between Cq derived from different test methodologies. In Figure 3-25 the comparison of Cq, based on all dynamic tests with fixed gust frequencies, show a very good agreement. Similar
good agreement is found in Figure 3-26 where they are compared to the ramp
gust tests.
Risø-R-1555(EN)
27
The comparisons of Cq derived from gust tests and static torque measurements,
Figure 3-27 and Figure 3-28, show a very good agreement provided that the
gust test results are somewhat adjusted due to variability of the wind tunnel.
Tests were made at different times of the year, with different climatic conditions. Other uncertain parameters involved in the determination of Cq are the
rotor inertia for the gust tests and the torque balance calibration for the static
torque measurements. Adjustments carried out for the test with the Risø anemometer were a shift in speed ratio by -0.006 and a scale up of Cq by 7%. Corresponding adjustments for the Thies tests were a shift in speed ratio by +0.007
and a scale up of Cq by 6%.
The reason for the marginal speed ratio shift has not been investigated in detail
but could be caused by an inherent filtering by the methods themselves. An increase in Cq value corresponds to an increase of the rotor inertia. So the need to
scale up the Cq values could be caused by an imprecise determination of the
inertia of the rotating parts. It can be speculated if the measurements of the rotor
inertia in still air may need to be corrected with the added mass due to accompanying mass of air attached to the cup surfaces. Neither the speed ratio shift
nor the Cq scale has any significant influence on the results from calculations of
dynamic behaviour. A scale up of Cq corresponds to an increased torque available for acceleration of the rotation but also an increased (negative) torque
available for deceleration of the rotation. It will be demonstrated later in this
report that it is the details around the equilibrium, i.e. the balance between the
two sides that counts.
The results of comparing the different types of torque measurements can therefore be considered very satisfactory. It is shown that it is possible to determine
the torque characteristics by an indirect way, i.e. by means of gust tests. This
has certain advantages, such as:
o
o
o
o
No modification of the anemometer is necessary
No extra components are necessary which can cause flow interference
Test can easily be performed with inclined flow (tilting the cup anemometer)
No balance equipment is necessary
One requirement is, of course, that access to a wind tunnel equipped with gust
facilities exists and that accurate measurements of the instantaneous wind speed
can be made.
3.8 Combined tilt and ramp-gust tests
In order to investigate the dynamic overspeeding for inclined flow, separate
tests were carried out with combinations of inclined flow at 21 fixed tilt angles
between ±47° and with gusts. The tests were carried out at 11.6 m/s tunnel
speed and gust amplitudes giving approx. 32% “turbulence intensity”. The gust
frequency was ramped up and down from 0-1.7-0-0.8-0 Hz during 210 seconds.
Figure 3-29 shows typical time traces from the test. Figure 3-30 shows a range
of torque coefficient curves derived by ramp-gust tests for the Risø cup anemometer. The torque coefficient curves can be assumed to contain essentially
all necessary information that is needed on the aerodynamic forces on the cup
anemometer rotor to predict the behaviour in arbitrary wind conditions.
28
Risø-R-1555(EN)
anem_193058_04_04_1
props_193058_04_04_1
13
12
Wind Speed [m/s]
11
10
9
8
7
6
5
4
3
0
30
60
90
120
150
180
210
Time [sec]
Figure 3-29 Time traces from the ramp-gust test at 0 degrees tilt angle. The
graph shows the average of the four propeller signals and the cup anemometer
signal from the whole 210 seconds sequence.
cq_-47.25_212935_02_04_1
0,4
cq_-38.38_212505_02_04_1
cq_-29.57_211926_02_04_1
cq_-23.57_211527_02_04_1
0,3
cq_-17.74_211123_02_04_1
cq_-14.80_210723_02_04_1
0,2
cq_-11.90_210326_02_04_1
cq_ -8.85_205920_02_04_1
0,1
cq_ -5.92_205521_02_04_1
cq_ -2.96_205120_02_04_1
0,0
cq_ -0.01_213419_02_04_1
cq_ 2.94_213839_02_04_1
-0,1
cq_ 5.91_214235_02_04_1
cq_ 8.91_215616_02_04_1
cq_ 11.85_220023_02_04_1
-0,2
cq_ 14.71_220516_02_04_1
cq_ 17.70_220922_02_04_1
-0,3
cq_ 23.59_221351_02_04_1
cq_ 29.55_221757_02_04_1
-0,4
cq_ 29.55_221757_02_04_1
cq_ 29.55_221757_02_04_1
-0,5
cq_ 38.46_222222_02_04_1
0,2 0,21 0,22 0,23 0,24 0,25 0,26 0,27 0,28 0,29 0,3 0,31 0,32 0,33 0,34 0,35 0,36 0,37 0,38
cq_ 47.25_222628_02_04_1
Figure 3-30 Torque coefficient curves Cq as function of speed ratio and tilt
angle for the Risø cup anemometer. The curves were derived from combined tilt
and ramp-gust-tests at 21 fixed inflow angles. The actual angle from each tilt
setting appears in the label.
3.9 Comparison of dynamic torque measurements
with angular response measurements
Obvious information, derived from the curves in the graph of Figure 3-30, are
the interception points between each torque coefficient curve and the Cq=0-line.
These speed ratios at various inflow angles can be normalised with the speed
ratio for horizontal flow and compared with the angular response from the static
tilt measurements. Figure 3-31 shows a typical comparison for a Vector cup
anemometer [8].
Risø-R-1555(EN)
29
Tilt tests with Vector A100LK, S/N 1476, Rotor R30K S/N M3T, 050629
1,05
cos
VEC_11_norm
VEC_8_norm
VEC_8R_norm
VEC_5_norm
tsr-0-aver (4,4,1)
Relative Speed
1,00
0,95
0,90
0,85
0,80
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt angle (neg from below)
Figure 3-31 Comparison of the results from the tilt tests and the torque interception points (black diamond symbols) from the ramp-gust tests for the Vector
A 100LK cup anemometer.
As can be seen from Figure 3-31 the agreement between the two types of measurements are, in this typical example, very good and convincing.
3.10 Gust Run Measurements at Horizontal Flow
Gust tests are carried out with the anemometer positioned in vertical position
and exposed to sinusoidal fluctuating wind with a mean wind speed of 8 m/s
and corresponding turbulence intensities of approximately 12, 16, 22 and 33%.
Gust frequencies are varied from 0.018Hz (60 second cycle) up to 1.7Hz. The
cup anemometer response and instantaneous wind speed from the propeller anemometers are recorded with high accuracy. This type of measurements give an
indication of the tendency to overspeed under pure horizontal gust conditions,
which also can be compared with the calculated response based on the measured
torque coefficient curves. Gust run measurements (overspeeding measurements)
are shown for a Vector cup anemometer in Figure 3-32, for a Risø cup anemometer in Figure 3-33, and for a classical Thies cup anemometer in Figure
3-34.
30
Risø-R-1555(EN)
Gust tests with Vector
1,10
VEC-tu%=31,7
VEC-tu%=22,0
VEC-tu%=16,1
VEC-tu%=11,4
Relative Speed
1,08
1,06
1,04
1,02
1,00
0,98
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
1,8
Gust frequency
Figure 3-32 Result from gust tests with a Vector cup anemometer. The graph
shows the ratio between the mean wind speed as seen by the anemometer and
the average of the propellers. The highest over-speeding is about 9%.
Gust tests with RISÖ P2546-Acc1
1,10
Relative Speed
1,08
RIS-tu%=31,6
RIS-tu%=21,9
RIS-tu%=16,1
RIS-tu%=11,2
1,06
1,04
1,02
1,00
0,98
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
1,8
Gust frequency
Figure 3-33 Result from gust tests with the Risø cup anemometer. The graph
shows the ratio between the mean wind speed as seen by the anemometer and
the average of the propellers. The highest over-speeding is about 6%.
Risø-R-1555(EN)
31
Gust tests with the Classical Thies Cup Anemometer
1,14
1,12
THC-tu%=31,6
THC-tu%=22,0
THC-tu%=16,1
THC-tu%=11,3
Relative Speed
1,10
1,08
1,06
1,04
1,02
1,00
0,98
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
1,8
Gust Frequency
Figure 3-34 Result from gust tests with the classical Thies cup anemometer.
The graph shows the ratio between the mean wind speed as seen by the anemometer and the average of the propellers. The highest over-speeding is about
12%.
3.11 Step Response Measurements
A step response test method is described in [15]. It is carried out by exposing
the anemometer to constant wind speed (UTARGET) and repeatedly stop and release the rotor and measure the pulse train from the anemometer. This exercise
is carried out several times for wind speeds at 5, 8 & 11m/s.
The distance constant (Dist) is determined from the pulse train by fitting an exponential function U(t) to the measured data, were t is measured from each time
of release of the anemometer, see Figure 3-35.
U (t ) = UTARGET (1 − exp
−t
τ
)
The distance constant is then given by:
Dist = U TARGET ⋅ τ
32
Risø-R-1555(EN)
9
Wind Speed [m/s]
8
7
ane_203104
fit_203104_17
6
5
4
3
10
11
12
13
14
Time [seconds]
Figure 3-35 The graph shows the response from a typical step response test at
8 m/s wind speed. From the graph each pulse from the anemometer can be seen.
The variation in rotational speed due to the three cups is clearly visible. The red
curve is fitted to the pulses to establish the time constant and the distance constant.
Table 3-3 shows the determined distance constants from a number of tests. The
distance constant varies randomly, due also to the position of release, between
1.1m and 1.6m, with an average of 1.346m and a standard deviation of 0.13m.
Table 3-3 Step response measurements of a cup anemometer
file
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
050901_202530
step
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
time
9,016
13,739
18,621
23,665
29,316
34,722
39,504
45,045
51,344
56,815
63,813
70,166
76,011
81,503
86,910
target
5,173
5,135
5,215
5,137
5,148
5,160
5,125
5,170
5,182
5,152
5,228
5,238
5,108
5,084
5,142
dist
1,199
1,458
1,476
1,209
1,557
1,309
1,417
1,273
1,567
1,282
1,477
1,455
1,323
1,143
1,152
fiterror
3,567
2,423
3,099
3,520
4,423
3,056
3,360
2,283
2,985
2,415
2,251
2,998
2,557
2,508
2,967
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
050901_203104
17
18
19
20
21
22
23
24
25
26
10,695
18,041
24,446
31,167
38,957
46,383
53,374
60,832
67,485
75,153
8,060
8,124
8,051
8,111
8,074
8,137
8,067
8,097
8,089
8,114
1,376
1,158
1,238
1,179
1,423
1,323
1,229
1,418
1,368
1,401
9,911
11,839
13,296
10,598
10,140
13,956
11,262
13,636
15,371
13,080
Risø-R-1555(EN)
average
std
1,353
0,143
33
050901_203104
050901_203104
050901_203104
27
28
29
83,593
90,339
97,022
8,088
8,177
8,096
1,541
1,217
1,453
10,842
15,974
9,059
1,333
0,118
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
050901_203327
30
35
36
38
39
40
41
42
43
44
45
10,784
17,018
23,274
30,501
37,054
43,792
50,475
57,616
63,991
70,209
77,544
11,196
11,302
11,322
11,151
11,254
11,230
11,214
11,266
11,188
11,148
11,205
1,280
1,216
1,292
1,094
1,421
1,581
1,505
1,451
1,419
1,317
1,305
26,436
28,910
31,310
28,683
31,722
34,161
38,492
32,915
28,768
30,331
26,900
1,353
0,139
all
1,346
0,131
3.12 Bearing Friction Measurements
Bearing friction measurements have been described in [5,14]. Friction in bearings of cup anemometers adds a small negative torque on the rotor which reduces the rotational speed a small amount. The friction depends on the design
and lubrication of the bearings. Often friction increases significantly at lower
temperatures due to higher viscosity of the lubrication substance. Friction in
bearings can be measured with a flywheel test. The rotor of the cup anemometer
is dismantled from the shaft, and a flywheel is fitted. The flywheel should have
the same weight as the rotor, so that the loading of the bearings can be considered the same. The flywheel should also have a sufficiently high inertia, so that
the rotor will keep spinning for a reasonably long time, a minute or more, for
the best conditions. It is important that the deceleration is not too fast, since this
will reduce the measurement sensitivity.
Friction must be measured under constant temperature conditions. Such conditions can be made in a closed climate chamber, where the rotor is excited by a
remote mechanism. An example of such an arrangement is shown in Figure
3-36, where the excitation is applied through a motor, driving a wheel with a
rubber ring, and applied to the flywheel by shortly activating an electric magnet.
34
Risø-R-1555(EN)
Figure 3-36 Bearing friction measurement setup in climate chamber. At left
heater, fan, sensors, excitation mechanism, and the mounting of the cup anemometer with the flywheel is show. At right the flywheel and the driving motor
is shown in detail.
The motor is set to rotate the cup anemometer at a rotational speed that corresponds to 20-25m/s. On excitation the motor will accelerate the flywheel to this
rotational speed, and on release, the flywheel will decelerate due to friction in
bearings and air friction on the flywheel.
The deceleration can be described with the differential equation:
I
dω
= − F (ω ) − 0.616πρ R 4 (νω 3 )1/ 2
dt
Here F (ω ) is friction in bearings as function of angular speed. The second term
on the right hand side is the air friction of the flywheel with the radius R [16].
The term is valid for laminar conditions, which is the case for Reynolds numbers below 300.000. For a flywheel of radius 3cm and rotation at 40rad/s, corresponding to 8m/s, the Reynolds number is 2400. Laminar conditions are therefore assumed for all cases during the tests. Table 3-4 lists air data to be used to
derive the flywheel friction. It is seen, that although both the air density and the
kinematic viscosity changes significantly with air temperature, the formula constant can in general be set to 0.009kg/m2s1/2 for the temperatures between 40°C
and -20°C.
Table 3-4 Air data for flywheel friction calculations
Air temperature
[°C]
-20
-10
0
10
20
40
Risø-R-1555(EN)
Air density ρ
[kg/m3]
1.39
1.35
1.29
1.25
1.21
1.12
Kinematic viscosity
ν [m2/s]
11.2*10-6
12.0*10-6
13.0*10-6
13.9*10-6
14.9*10-6
17.1*10-6
0.616*π*ρ*ν1/2
[kg/m2s1/2]
0.009002
0.00905
0.009001
0.009019
0.009039
0.008963
35
The differential equation can be reduced to:
I
dω
= − F (ω ) − 0.009kg / m 2 s1/ 2 ⋅ R 4 (ω 3 )1/ 2
dt
Isolating the bearing friction term, one gets:
F (ω ) = − I
dω
− 0.009kg / m 2 s1/ 2 ⋅ R 4 (ω 3 )1/ 2
dt
The friction measurement in the climate chamber gives a deceleration curve of
angular speed as function of time. This curve is fitted to a 3rd order polynomial,
which can easily be differentiated to a 2nd order polynomial. The data are fitted
in the interval 4-16m/s only. The friction within the 4-16m/s wind speed interval is again fitted to a second order polynomial:
F (ω ) = f1 + f 2ω + f3ω 2
There may be some torque associated with the sensing of the rotational speed of
the cup anemometer, which should be included in the bearing friction measurement. The cup anemometer should therefore be electrically connected for measurements during the tests.
Figure 3-37 shows a typical deceleration run with a cup anemometer with two
pulses per rev. Such pulses are seldom opposed exactly by 180º, which is indicated by the spreading of data during the deceleration. Figure 3-38 shows the
total friction, the flywheel friction, and the deduced bearing friction. Figure
3-39 and Figure 3-40 show measured decelerations and fitted friction of five
successive tests at -18°C. The results show good reproducibility. The cup anemometer was a RISØ P2546 cup anemometer, see [14]. Flywheel characteristics
were: weight 51g, radius 0,030m, inertia 2,30*10-5 kgm2. Figure 3-41 show
typical measured friction at temperatures form 40°C to -20°C.
Risø deceleration (2 pulses/rev)
100
90
Run
3rd order polynomia fit
80
Angular speed (rad/s)
70
60
50
40
30
20
10
0
0
5
10
15
20
25
30
35
40
45
50
55
Time (sec)
Figure 3-37 Typical deceleration run with a Risø cup anemometer with two
pulses per rev at 20ºC
36
Risø-R-1555(EN)
Risø deceleration (2 pulses/rev)
4.00E-05
3.50E-05
Friction kg*m^2/s^2
3.00E-05
Torque
Flywheel
Friction
2.50E-05
2.00E-05
1.50E-05
1.00E-05
5.00E-06
0.00E+00
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
Angular speed (rad/s)
Figure 3-38 Total friction, flywheel friction and bearing friction of Risø cup
anemometer as function of angular speed at 20ºC
RISØ P2546 cup anemometer - Flywheel tests Reproducability
90
80
Angular speed [rad/s]
70
-18deg
-18deg
-18deg
-18deg
-18deg
60
50
40
30
20
10
1
2
3
4
5
6
7
8
Time [s]
Figure 3-39 Five successive flywheel deceleration tests of RISØ P2546 cup anemometer
Risø-R-1555(EN)
37
RISØ P2546 cup anemometer - Flywheel tests
6,0E-04
Friction Torque [Nm]
5,0E-04
4,0E-04
-18deg
-18deg
-18deg
-18deg
-18deg
3,0E-04
2,0E-04
1,0E-04
0,0E+00
10
20
30
40
50
60
70
80
90
Angular Speed [rad/s]
Figure 3-40 Five successive friction torque measurements of Risø cup anemometer
RISØ P2546 cup anemometer - Flywheel tests
6,0E-04
40deg
35deg
30deg
25deg
20deg
15deg
10deg
8deg
6deg
4deg
2deg
0deg
-2deg
-4deg
-6deg
-8deg
-10deg
-12deg
-14deg
-16deg
-18deg
-20deg
Friction Torque [Nm]
5,0E-04
4,0E-04
3,0E-04
2,0E-04
1,0E-04
0,0E+00
10
20
30
40
50
60
70
80
90
Angular Speed [rad/s]
Figure 3-41 Friction measurements of Risø cup anemometer [14]
38
Risø-R-1555(EN)
3.13 Rotor Inertia Measurements
The cup anemometer rotor inertia is measured by a simple oscillation method in
which the rotor is set to oscillate around its axis, see Figure 3-42.
Figure 3-42 Determination of moment of inertia by oscillations of a Vector cup
anemometer rotor. Length of strings 0.8m.
The rotor is set to oscillate with a not too high amplitude as the air damping will
influence on the cups. A number of oscillations is made, and the average time of
one oscillation determined.
From the oscillation tests, the inertia can be found from the formula:
I=
T 2 Mgr 2
4π 2 l
where:
T
M
r
l
g
is average time of one oscillation
is mass of rotor
is radius from axis of rotation to the three strings
is the length of the strings
is gravitational acceleration ~9.81m/s2
Risø-R-1555(EN)
39
4. Time Domain Modelling of CupAnemometers
4.1 Introduction
Modelling of cup anemometers is known since 1929, where Schrenk [17]
measured torque on cup anemometers to analyse dynamic characteristics. Several
scientists have since then made modelling of cup anemometers, and
improvements have been made with better insight in the physics of the
instruments. Two time domain models are presented in the following chapters.
The use of the models in simulation for classification of five cup anemometers is
presented in report [8].
The first model was developed in the CLASSCUP project [5]. It is in this report
referenced as the Tilt-Response&Torque-Coefficient model (TRTC). It is based
on a measured tilt response, a measured torque coefficient curve for horizontal
flow, and measured bearing friction curves. This model is also described in annex
J of the IEC standard [7].
The second model, here referenced as the Inclined-Flow-Torque-Coefficient
model (IFTC), is based on measured torque coefficient curves for a range of
inclined flow angles. This new model is developed in the ACCUWIND project,
and is only available due to the possibility to measure torque coefficient curves
under inclined flow conditions.
The cup anemometer rotor torque is in both models calculated from instantaneous
3D wind time traces, derived by artificial wind generators representing turbulent
field wind conditions, see chapter 5.
4.2 Generalization of the rotor torque
Normalization of the rotor torque was originally proposed by Schrenk [17]. He
normalized measured torque to parabolas. Normalization of rotor torque to one
torque coefficient curve was also demonstrated in the CLASSCUP project [5].
The torque on one cup of a RISØ cup anemometer was measured in the FFA LT5
wind tunnel for varying angular speed and varying wind speed. The torque
measurements are shown in Figure 4-1 and Figure 4-2.
40
Risø-R-1555(EN)
Measured Torque Characteristics for one RISØ cup
5
4
Torque (mNm)
3
Ome=25 rad/s
Ome=40 rad/s
Ome=55 rad/s
Poly. (Ome=25 rad/s)
Poly. (Ome=40 rad/s)
Poly. (Ome=55 rad/s)
2
1
0
-1
-2
-3
-4
0
2
4
6
8
10
12
Wind speed (m/s)
Figure 4-1 Measured torque curves for RISØ cup anemometer (one cup) with
fixed rotational speed and varying wind tunnel speeds
Measured Torque Characteristics for one RISØ cup
6
Torque (mNm)
4
Wsp=5 m/s
Wsp=8m/s
Wsp=11m/s
Poly. (Wsp=5 m/s)
Poly. (Wsp=8m/s)
Poly. (Wsp=11m/s)
2
0
-2
-4
-6
0
20
40
60
80
Angular speed (rad/s)
Figure 4-2 Measured torque curves for RISØ cup anemometer (one cup) with
fixed wind tunnel speed and varying rotational speeds
The measured torque curves of Figure 4-1 and Figure 4-2 can be generalized
nicely to one normalized torque coefficient curve, where the torque coefficient is
defined as:
CQA =
QA
1
ρ ARU 2
2
Q A is rotor torque, i.e. in this example three times one cup torque. The measured
torque should be subtracted bearing friction torque before normalization in
clearing that the torque coefficient represents pure rotor aerodynamics.
Risø-R-1555(EN)
41
The speed ratio, being the x-axis in the normalized torque coefficient curve, is
defined by:
λ=
ωR
U − Ut
The wind speed U is subtracted a threshold wind speed U t which corresponds to
most of the offset in the calibration line. The rest and small part of the offset is
due to friction. The introduction of a threshold wind speed is a necessity that
makes the measured and normalized torque coefficient curve consistent with the
static calibration expression at zero torque. In the case of no friction the speed
ratio can be transformed into a calibration expression:
λ=
ωR
U −Ut
⇔U =
ωR
λ
+ U t ⇔ ω = (U − U t )
λ
R
All measured data from Figure 4-1 and Figure 4-2 are normalized and shown in
Figure 4-3. The figure shows that all data collapse into one curve. The measured
data are seen to fit reasonably well to a parabola. It should be mentioned though,
that parabolas do not describe the important details of the torque curve, which are
necessary for realistic simulation of cup anemometer dynamics.
RISØ Normalised Torque Coefficient
2.0
Ome=25
1.0
Ome=40
0.0
Ome=55
U=5
-2.0
U=8
Cq
-1.0
U=11
-3.0
U=8, Tu=16, f=2
-4.0
Parabola fit
-5.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Speed ratio
Figure 4-3 Normalized torque coefficient of the RISØ cup anemometer for
various wind speeds and angular speeds
4.3 The Tilt-Response & Torque-Coefficient
Model (TRTC)
The TRTC model assumes the forces on the cup anemometer rotor to be based
on a torque, due to aerodynamic forces on the rotor and a friction torque in bearings, due to lubrication and contact of moving surfaces. The aerodynamic
torque is dependent on wind speed, angular speed of the rotor and flow inclination angle, and the friction torque is dependent on temperature and angular
speed of the rotor, so the total torque can be expressed as:
Q = Q A (U , ω, α ) − QF (T , ω )
Q A is the aerodynamic torque on the rotor and QF is the friction torque in
bearings, a positive quantity. The aerodynamic torque is expressed in terms of
42
Risø-R-1555(EN)
the angular response curve and the normalized torque coefficient curve for horizontal flow conditions, which are combined to give the total aerodynamic
torque.
Angular response
The instantaneous wind vector is reduced to an equivalent horizontal wind
speed, which is assumed to generate the same aerodynamic torque. First, the
instantaneous wind vector for the time t is determined from the artificial wind
generator:
r
U = ( u , v , w)
Where u, v, w are the longitudinal, the transversal and the vertical wind speed
components respectively. The instantaneous flow inclination angle is:
α = A tan
w
u + v2
2
The influence of the flow inclination angle is expressed in the angular response
curve of the cup-anemometer, and is dependent on flow inclination angle and
wind speed, in this case the scalar of the wind vector is substituted with the
constant wind tunnel wind speed:
r
U vec = U = u 2 + v 2 + w2
The equivalent horizontal flow wind speed U eq is derived from the scalar wind
vector by multiplication with the angular response Fα (U vec , α ) :
U eq = Fα (U vec , α ) ⋅ U vec
The angular response is linearly interpolated between, at first wind speed curves,
then flow angle.
Aerodynamic torque for horizontal flow
For the horizontal flow conditions with the equivalent horizontal flow wind
speed, the aerodynamic torque is derived from the torque coefficient curve.
First, the speed ratio is determined:
λ=
ωR
U eq − U t
The torque coefficient curve CQA (λ ) can either be fitted to a higher order
polynomial, forced through the equilibrium speed ratio (speed ratio at zero
torque) with a higher order (for instance 11) or be tabulated. The aerodynamic
torque is determined by:
QA =
1
ρ ARU 2CQA (λ )
2
Bearing friction
Bearing friction is linearly interpolated between parabolic friction curves at different temperatures:
QF = f 0 (T ) + f1 (T ) ⋅ ω + f 2 (T ) ⋅ ω 2
Where f 0 , f1 , f 2 are constants for given temperatures.
Total driving torque
The driving torque exerts acceleration on the rotor according to the differential
equation:
Risø-R-1555(EN)
43
I
dω
= Q = Q A − QF
dt
In numerical calculations, the acceleration is assumed constant in a short time
step t , and a new angular speed is calculated with the Euler method:
Δω =
Q
Δt
I
When the angular speed is known, the indicated (measured) wind speed of the
cup anemometer is found by applying the calibration expression:
U = Acalω + Bcal
This wind speed can then be compared to the “true” wind speed, which in the
IEC standard, ref. 8, is defined as the horizontal wind speed:
U hor = u 2 + v 2
Deviations, calculated for classification purposes are based on simulation of
measurement, often using 10min averaging:
U dev =
∑U− ∑U
10 min
hor
10 min
Calculations, based on the TRTC model are shown in [5,8,14].
Fitting of data to the cup anemometer model
In practice, not all characteristics on a cup anemometer can be made under the
same conditions. The use of different testing institutes, different wind tunnels,
and the variation of atmospheric conditions as temperature and humidity may
influence on the measurement results, so a fitting of the data to the model is
necessary, for instance as shown in [14, chapter 4.2]. The calibration curve,
preferably based on an accredited calibration certificate, is the basis for the fitting process. The process is accomplished by the following steps:
1. Normal calibration in wind tunnel at certain calibration temperature to
generate calibration curve
2. Measure angular response curves in wind tunnel
3. Measure torque curves in wind tunnel at certain temperature
4. Measurement of bearing friction in climate chamber
5. Determine bearing friction at temperature for torque curve measurements
6. Add bearing friction to measured torque curves
7. Determine aerodynamic torque coefficient curve
8. Fit aerodynamic torque coefficient curve to linear calibration curve by
numerical simulation of calibration so that threshold wind speed and a
torque correction value is found
The last point is very important. It is absolutely essential that simulation with
the cup anemometer model of the linear calibration is made and fitted successfully if the simulations shall be used for classification purposes. Successfully in
this context means that the simulations shall almost not deviate from the calibration line and at least all deviations shall be much less than the individual
variations of the calibration points from the line.
44
Risø-R-1555(EN)
4.4 The Inclined-Flow-Torque-Coefficient Model
(IFTC)
The IFTC model is in many aspects similar to the TRTC model, but deviates in
the way aerodynamic torque is simulated. The aerodynamic torque is, as for the
TRTC model, dependent on the wind speed, angular speed of the rotor and flow
inclination angle, but the IFTC model uses a more direct approach. Once the
normalised torque coefficient curves are known for horizontal flow as well as
for inclined flow they can be used to find the response of the anemometer when
it is exposed to artificial wind conditions including wind variations in all directions. The flow inclination angle is used directly to determine the torque coefficient, and no angular response curve is used.
The simulation procedure in using the IFTC model here is as follows:
Assume friction is insignificant and that the anemometer rotates with an angular
speed ω and that wind data (artificial or real) including all three velocity components are known with a high time resolution. (usually 20 Hz or more), then
for each time step:
1.
Determine the scalar wind speed and inflow angle
2.
Determine the speed ratio and the torque coefficient CQ (λ , α )
3.
4.
5.
6.
from the database
Calculate the torque value
Calculate the angular acceleration (deceleration)
Calculate the angular speed ω in the next time step
Calculate the wind speed indicated by the anemometer
The inflow angle
The scalar vector wind speed is calculated as the resultant of the three wind velocity components:
U vec = u 2 + v 2 + w2
The inflow angle is calculated as the angle between the vertical and the horizontal components.
α = A tan
Aerodynamic torque at inclined flow
The speed ratio is calculated as:
λ=
w
u2 + v2
ωR
U vec − U t
The torque coefficient for the present time step is interpolated from the database
of torque coefficient curves CQ (λ , α ) , shown in Figure 3-30. Linear interpolation is used to establish a CQ (α ) table for the present speed ratio. The CQ
value for the present inflow angle is derived from the table by cubic spline interpolation.
The driving torque can now be calculated according to:
Q=
Risø-R-1555(EN)
1
2
ρ ARU vec
Cq
2
45
Determination of measured wind speed
The angular acceleration is calculated according to:
dω
Q
= ω& =
dt
I
This acceleration is assumed to be constant and so the angular speed at the next
time step is given by:
ωt +Δt = ωt + ω& ⋅ Δt
When the rotational speed of the anemometer is known, the wind speed indicated by the anemometer is calculated using the standard calibration constants.
The use of the IFTC model on five cup anemometers is shown in [8].
5. External Conditions for Cup
Anemometer Classification
5.1 External Operational Conditions
The conditions for classification of cup anemometers in this report refer to the
classification system presented in the IEC standard, ref. 8. The standard has two
classification categories for cup anemometers. The first, category A, relates to
“ideal” sites, which consist of reasonably flat terrain, where the terrain influence
is small on power curve measurements. The second, category B, relates to complex sites that do not meet the requirements of an “ideal” or reasonably flat site,
and where the terrain influences significantly on power curve measurements.
Table 5-1 show the operational ranges required for the classification categories
in the IEC standard. The table includes information on the wind spectra being
used in this report with 350m longitudinal length scale for the Kaimal spectrum
for Class A, and 170m length scale for the isotropic von Karman spectrum for
Class B.
Table 5-1 Operational ranges of Class A and Class B category classification
46
Risø-R-1555(EN)
Wind speed range to
cover [m/s]
Turbulence intensity
Turbulence structure
σu/σv/σw
Air temp. [°C]
Air density [kg/m3]
Average flow inclination angle [°]
Class A
Terrain meets requirements in annex B of standard
Min
Max
4
16
0,03
0,12+0,48/V
1/0,8/0,5
(non-isotropic turbulence)
Kaimal wind spectrum
with a longitudinal turbulence length scale of
350m
0
40
0,9
1,35
-3
3
Class B
Terrain does not meet requirements in annex B of
standard
Min
Max
4
16
0,03
0,12+0,96/V
1/1/1
(isotropic turbulence)
Von Karman wind spectrum with a longitudinal
turbulence length scale of
170m
-10
40
0,9
1,35
-15
15
5.2 Wind speed range
Most wind speed measurements in wind energy applications are related to wind
resource and wind utilisation purposes. For these purposes 10 minute averages
are used. The IEC standard [7], and Table 5-1 requires the 10 minute average
wind speed range 4-16m/s for classification purposes. Below 4 m/s the power in
the wind is insignificant and above 16 m/s, the power from the wind turbine is
regulated to be almost constant and independent of the wind speed, so the 416m/s range is appropriate for this analysis.
5.3 Turbulence ranges
5.3.1
Turbulence intensity levels
The IEC safety standard [18], gave guidance as to the relevant range of turbulence intensities for design of wind turbines. The IEC standard on performance
measurements [7], uses similar expressions for the turbulence ranges. The maximum standard deviations of the wind speed for Class A requirements are:
σ 1 = 0.48m / s + 0,12Vhub ⇒ Ti =
0.48m / s
+ 0,12
Vhub
The equivalent requirements for class B are:
σ 1 = 0.96m / s + 0,12Vhub ⇒ Ti =
0,96m / s
+ 0,12
Vhub
The maximum turbulence for class A requirements are thus 24% at 4m/s and
15% at 16m/s, and for class B 36% at 4m/s and 18% at 16m/s, respectively. The
lowest turbulence intensities that are considered in the IEC standard [7] are 3%
at all wind speeds. Figure 5-1 shows the turbulence intensity ranges for class A
and class B requirements.
Risø-R-1555(EN)
47
0.4
Max turbulence Class A
Max turbulence Class B
Min turbulence
0.35
Turbulence intensity
0.3
0.25
0.2
0.15
0.1
0.05
0
0
2
4
6
8
10
12
14
16
18
Wind speed (m/s)
Figure 5-1 Turbulence intensity ranges as function of wind speed for class A
and class B requirements
5.3.2
Turbulence spectra
The three-dimensional turbulent wind spectra are site specific, and do seldom
follow the general average spectra, but for classification purposes it is appropriate to apply standard spectra. The spectra being applied in this IEC safety standard [18] are the Kaimal spectrum model for non-isotropic conditions and the
von Karman spectrum model for isotropic conditions. These spectra are also
applied here.
The Kaimal spectra of the three turbulence components, used for class A cup
classification, are modelled by:
fSk ( f )
σ
2
k
=
4 fLk / U
(1 + 6 fLk / U )5 3
The average wind speed U , the standard deviations σk and the length scales Lk,
where the k index indicate the longitudinal, transversal and vertical turbulence
components are parameters for which their ranges must be determined. The relation between the standard deviations and the length scales in the spectra are
shown in Table 5-2.
Table 5-2 Standard deviations and length scales for the three turbulence component Kaimal spectra
Standard deviation σ k
Integral length scale Lk
Longitudinal
Transversal
Vertical
L1
0.33 ⋅ L1
0.08 ⋅ L1
σ1
0.8 ⋅ σ 1
0.5 ⋅ σ 1
In the IEC safety standard [18], the longitudinal length scale L1 for heights
above 30m is determined as 170m, while it is required to be set equal to 350m
in the IEC power performance measurement standard [7]. This is a factor 2,3
times higher.
48
Risø-R-1555(EN)
The average length scale of the longitudinal spectrum is varying over a broad
spectrum. For a specific site at Vindeby [19], the longitudinal length scales have
been analysed for both onshore (LM) and offshore positions (SMS and SMW),
as shown in Figure 5-2. The dashed curves are for 3m height.
Figure 5-2 Longitudinal length scales measured onshore and offshore at Vindeby [19].
The longitudinal length scale of 350m shows good agreement with the requirement measured average length scales in Figure 5-2.
In the IEC power performance standard [7] it is required to use the Kaimal spectrum. For clarity, the Kaimal spectrum is defined for non-isotropic turbulence
while the von Karman is defined for isotropic turbulence conditions. The von
Karman is for this work used for Class B classification with a length scale of
170m, as shown in Table 5-1. This length scale is equivalent to the 350m length
scale of the Kaimal spectrum.
Risø-R-1555(EN)
49
The von Karman isotropic turbulence spectrum, used for class B cup classification in this study, being the same for all three turbulence components, is modelled by:
fSk ( f )
σ
2
k
=
4 fLk / U
(1 + 71( fLk / U ) 2 )5 6
In the IEC safety standard [18], the longitudinal length scale L1 for a von Karman spectrum for heights above 30m is determined to be 74m. Using the same
relative length scale factor of 2.3 for the Kaimal spectrum, as found for the von
Karman spectrum, the length scale is set to 170m.
5.4 Air temperature ranges
The environmental air temperature is the temperature to which the cupanemometer is exposed, but not necessarily the temperature of the bearings,
which might be heated in several types of cup-anemometers. The air temperature ranges of operational wind turbines are according to the IEC safety code
[18] -10°C to 40°C for normal conditions and -20°C to 50°C for extreme conditions. The temperature range -10°C to 40°C corresponds to the requirements of
Class B classification in Table 5-2. Though the table include temperatures down
to -10°C the category does not specifically consider conditions with rime, snow
and ice, and such conditions are not taken into account in the present analysis.
The Class A category is avoiding rime, snow and icing conditions as the temperature range is limited to 0°C to 40°C.
5.5 Air density range
The air density range is in Table 5-2 required to vary between 0.90 to 1.35
kg/m3. The air density is affecting the ratio of frictional forces to aerodynamic
forces on cup anemometers. This ratio becomes significant when temperatures
are low and lubricants increase viscosity or when cup anemometers have small
cups and rotor arms compared to size of bearings. For altitudes from sea level
up to 2000m the standard air density ranges from 1.225 to 1.006 kg/m3, according to ISO Standard Atmosphere [20]. For a constant altitude a typical air density range of ±10%, including air humidity variations, is normal. The air density
range for a power performance measurement is thus adequate for the classification.
5.6 Flow inclination angle ranges
During power performance measurements the flow inclination varies. For a flat
test site (Risø test site, Roskilde, Denmark) the flow inclination varies during a
measurement period as shown in Figure 5-3.
50
Risø-R-1555(EN)
Angle (deg) mean vertical wind speed to horiontal wind
speed)
25
20
15
10
5
0
-5
-10
-15
-20
-25
0
2
4
6
8
10
12
14
16
18
Horizontal wind speed (m/s)
Figure 5-3 Variations of flow inclination angle (mean vertical component to
mean horizontal component) for a flat site at 30m height (Risø, Roskilde, Denmark)
Figure 5-3 shows that measured flow inclination angles vary a little more than
the required range in Table 5-1 at low wind speeds for a Class A cup classification. Wind turbine sites and power performance measurements are performed
also in mountainous terrain where the slope of the terrain may be high. To cover
such applications, a range of flow inclination angles of -15° to 15° for class B in
Table 5-1 seems reasonable, as the terrain slopes themselves may be of this size.
5.7 Artificial 3D Wind Generators
In modelling cup anemometers it is important to know in detail the wind input.
It is thus very appropriate to use artificial 3D wind generators to generate wind
with the required characteristics as required in Table 5-1 for.
Several artificial wind generators are available. The wind generators used for
the present analysis are made by Jakob Mann [21]. They are made of two algorithms, one called Kaimal.exe, the other vonKarman.exe, generating wind from
the Kaimal and von Karman models, respectively.
The algorithms simulate turbulent atmospheric wind fields, and are based on a
model of the spectral tensor for atmospheric surface-layer turbulence at high
wind speeds They simulate three-dimensional fields of three components of the
wind velocity fluctuations. The spectral tensor has been compared with and adjusted to several spectral models used in wind engineering, for example the
Sandia method [22].
Risø-R-1555(EN)
51
6. Application of Classification Procedures
6.1 Classification of cup anemometers
The calibration methods, the two cup anemometer models, TRTC and IFTC,
and the external operational conditions according to the IEC performance measurement standard [7], and Table 5-1 have been applied to five different types of
cup anemometers for classification. The results are presented in the report ref. 8.
Results of the IFTC model are shown without influence of friction, and results
of the TRTC model are shown for angular response measurements made by
both DEWI and FOI.
6.2 Robustness of applied procedures
There are many steps in making a classification of a cup anemometer. First of
all, the calibrations: normal calibration, angular response calibration, torque
calibration and friction calibration have uncertainties. The calibrations deviate
in different methods being used, and in different institutes applying the procedures. As an example, variations in Round Robin tests of normal calibrations
between cup anemometer calibration institutes within the MEASNET organization are of the order of 1%.
Cup anemometer models vary. They vary with the parameters they take into
account and how they interpolate data, and they vary how well they simulate the
physical characteristics.
Robustness of classification procedures is reached when the application of the
procedures gives almost the same results for different institutes performing the
classification and for different approaches with sufficient capability being applied. This chapter will deal with some aspects of robustness of the procedures
being applied.
6.3 Calibration Robustness
6.3.1
Differences in normal calibrations
In the MEASNET organization web site [23] regular Round Robin tests are performed in order to keep track of the deviations between the cup anemometer
calibration institutes. The organization has rules that require all calibration institutes to perform calibrations within 1% of average. This requires the calibration
procedure in the IEC power performance measurement standard [7] being applied. In 1985 when a common cup anemometer calibration procedure was unavailable, a Round Robin test varied between calibration institutes by 6%.
52
Risø-R-1555(EN)
6.3.2
Differences in tilt response measurements
Measurements of static angular response of a range of cup anemometers have
been carried out by FOI and DEWI in the ACCUWIND project. The deviation
between these measurements can be regarded as a measure of the robustness of
one quantity used in the classification process namely the angular response. The
deviations between FOI and DEWI for the five examined anemometers are presented in Figure 6-1 to Figure 6-6.
FOI-DEWI tilt comparison
1,10
cos
FOI_NRG_8
DEWI_NRG_8
FOI_RIS_8
DEWI_RIS_8
FOI_THF_8
DEWI_THF_8
FOI_VAI_8
DEWI_VAI_8
FOI_VEC_8
DEWI_VEC_8
Relative Speed
1,05
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-1 Angular response measurements on five cup anemometers by FOI
and DEWI at 8m/s
FOI-DEWI tilt comparison
1,10
Relative Speed
1,05
cos
FOI_NRG_8
DEWI_NRG_8
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-2 Angular response measurements on NRG cup anemometer by FOI
and DEWI at 8m/s
Risø-R-1555(EN)
53
FOI-DEWI tilt comparison
1,10
Relative Speed
1,05
cos
FOI_RIS_8
DEWI_RIS_8
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-3 Angular response measurements on Risø cup anemometer by FOI
and DEWI at 8m/s
FOI-DEWI tilt comparison
1,10
cos
FOI_THF_8
DEWI_THF_8
Relative Speed
1,05
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-4 Angular response measurements on Thies First Class cup anemometer by FOI and DEWI at 8m/s
54
Risø-R-1555(EN)
FOI-DEWI tilt comparison
1,10
Relative Speed
1,05
cos
FOI_VAI_8
DEWI_VAI_8
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-5 Angular response measurements on Vaisala cup anemometer by
FOI and DEWI at 8m/s
FOI-DEWI tilt comparison
1,10
Relative Speed
1,05
cos
FOI_VEC_8
DEWI_VEC_8
1,00
0,95
0,90
0,85
-50
-40
-30
-20
-10
0
10
20
30
40
50
Tilt Angle (Neg=from below)
Figure 6-6 Angular response measurements on Vector cup anemometer by
FOI and DEWI at 8m/s
The causes for the deviations in the calibrations have not been examined in detail. FOI is using a closed test section and the anemometer is moved back and
forth in the test section to obtain different tilted positions. The wind tunnel used
by DEWI is an open test section and the anemometer rotor centre is kept fixed
for all tilted positions.
6.3.3
Differences due to inertia
As demonstrated in section 3.7.3 it was necessary to adjust the torque coefficient values used in the in-direct torque measurement method for both the Risø
and the Thies classical cup anemometers to match the direct measured normalized torque curves. The Cq values for the Risø anemometer were scaled up by
Risø-R-1555(EN)
55
7% and the Cq values for the Thies anemometer were scaled up by 6%. One of
the scale factors used in the formula to derive the torque curves is the cup inertia. In section 3.7.4 it was discussed if the differences could be caused by too
low cup inertia.
To demonstrate how an error in any of the scale factors influence the calculation
of the class-index a sensitivity calculation was performed. The inertia values of
the five commercial cup-anemometers were changed ±20% from the nominal
values. This gave rise to a maximum change in the class-index of less than 1%
for all anemometers.
This shows that the accuracy of the absolute values of the scale parameters are
of lesser importance. A scale up of Cq corresponds to an increased torque available for acceleration of the rotation but also an increased (negative) torque
available for deceleration of the rotation. It is the details around the equilibrium,
i.e. the balance between the two sides that counts.
6.4 Cup Anemometer Model Robustness
The most robust model of a cup anemometer is capable to simulate any controlled flow condition, especially all calibration flow conditions, where the calibration results are fitted to the model.
6.4.1
Simulation of normal calibration
The most important simulation, as mentioned in Chapter 4.3, is a simulation of
a normal calibration. A normal calibration is the best controlled flow condition
a cup anemometer can be exposed to. The model must be able to simulate the
calibration line very accurately, see Figure 6-7, where the TRTC model simulates a calibration line [14]. For cup anemometers quite sensitive to friction, this
calibration line will vary with temperature.
Deviation of calibration points from line
0,050
0,040
Deviation [m/s]
0,030
0,020
cali deviation
simu deviation
0,010
0,000
-0,010
-0,020
-0,030
0
2
4
6
8
10
12
14
16
18
Wind speed [m/s]
Figure 6-7 Deviations from calibration line for a calibration and a corresponding simulation of a Risø cup anemometer [14]
56
Risø-R-1555(EN)
6.4.2
Comparison of measured and calculated step response
Another well-controlled flow condition is a step response measurement; see
Figure 6-8, where the step response results from the measurements presented in
Chapter 3.11. The step response is compared with a corresponding calculated
step response by a simulation with the IFTC model. Figure 6-8 shows a quite
good fit of measured and calculated data. The measurement of a step response is
difficult, though, when number of pulses per rev is small.
9
Wind Speed (m/s)
8
7
Wind Speed (input)
Anemometer Response (output)
Response fit
6
Distance constant from fit: 1,45
5
4
3,5
4,0
4,5
5,0
5,5
6,0
Time (sec)
Figure 6-8 Calculated step response for an anemometer based on measured
CQ (λ ) and applied step change. The distance constant from the calculation
and fitted value 1.45 agrees rather well with the measured values 1.35±0.13 of
Chapter 3.11.
6.4.3
runs
Comparison of measured and calculated response from gust
Gust run measurements with pure horizontal flow presented in Chapter 3.10,
and demonstrated in Figure 3-32 to Figure 3-34, also represent reasonably well
controlled flow conditions. The dynamic response expressed as overspeeding
can be compared with corresponding calculated values using the IFTC model.
This comparison should give an indication on how well the procedure works
under pure horizontal conditions.
As input to the calculations the derived torque database CQ (λ , α ) and pure sinusoidal wind sequences were used. The wind sequences were generated with
wind variations only in the longitudinal component and amplitudes and gust
frequencies similar to the gust measurements (as seen by the propellers). Figure
6-9 shows a comparison for the classical Thies anemometer.
Risø-R-1555(EN)
57
Measured and Calculated Dynamic Effects for the Thies Classical Cup Anemometer
1,14
1,12
Meas, tu=33
Meas, tu=16
Calc, tu=32,20
Calc, tu=16,10
Speed Ratio
1,10
1,08
Meas, tu=22
Meas, tu=12
Calc, tu=22,00
Calc, tu=11,70
1,06
1,04
1,02
1,00
0,98
0,0
0,5
1,0
1,5
2,0
Gust Frequency
Figure 6-9 Comparison of measured and calculated response from gust tests
with the classical Thies cup anemometer.
The agreement in Figure 6-9 is very good at lower gust frequencies. The deviations at higher gust frequencies are caused by the fact that the tunnel can not
maintain the high amplitudes. The wind speed as indicated by the propellers
show a reduction in turbulence (amplitudes) at the highest frequency by approximately 3-8%. A more thorough comparison would be to use the actual
wind speed sequences from the propellers in the calculations instead of the generated constant amplitude sequences.
6.5 Robustness of Classification
Classification is robust when calibration procedures are robust, cup anemometer
models are robust, and when classification results are almost the same when
performed by different institutes on different examples of the same cup anemometer make and type. In this respect it must be mentioned that one type of
cup anemometer varies in calibration line in normal calibrations and over time,
and this is the reason why cup anemometers are calibrated separately, and before and after a power curve measurement. A spreading of classification results
should therefore also be expected.
A comparison of four classifications is made on a Risø type cup anemometer,
according to the IEC power performance standard [7], and Table 5-2. Classification 1 in Table 6-1 is made with the TRTC model [14], based on tilt and
torque measurements by FOI, normal calibration by DEWI, and friction and
inertia measurements by Risø. The external operational conditions in Table 5-1
were different from [7], so a re-calculation was made. Classifications 2, 3 and 4
in Table 6-1 are based on another cup anemometer of the same type. Classification 2 is also made with the TRTC model. Normal calibration, tilt and torque
measurements are made by FOI, and friction and inertia measurements are made
by Risø. Classification 3 is the same as 2, except that the tilt response measurement is made by DEWI, Figure 6-3. Classification 4 is made with the IFTC
model. Normal calibration, inclined flow torque coefficients and inertia are de-
58
Risø-R-1555(EN)
termined from inclined flow measurements by FOI. The results of the four classifications are shown in Table 6-1.
Table 6-1 Four classifications of Risø cup anemometer with variations in angular response
Classification no
Class A
Class B
Class A
Class B
Category
Category
Category
Category
Horizontal Horizontal
Vector
Vector
1) TRTC,
1.3
6.1
1.8
10.2
R-1364 report
2) TRTC, FOI
1.4
5.1
1.7
9.2
tilt response
3) TRTC, DEWI
1.9
8.0
2.4
12.0
tilt response
4) IFTC
1.3
5.0
1.7
9.2
no friction
The four classifications in Table 6-1 show reasonably good correlations, but
also with a specific deviation that need a comment. Classification 3 deviates
from the others in significantly higher class values for all categories. The reason
for this should be found in the angular response differences in Figure 6-3. The
DEWI angular response curve is for negative tilt angles significantly lower than
the FOI curve even for small tilt angles.
Another comparison of classifications is made focusing on the influence of friction. In this case classifications were made on a cup anemometer that has relatively high friction torque compared to aerodynamic torque. Classifications 5
and 6 in Table 6-2 correspond to the conditions of classification 2 in Table 6-1,
with and without friction. Classifications 7 and 8 correspond to the conditions
of classification 3, respectively, and conditions of 9 correspond to 4. All classifications are shown in Table 6-2.
Table 6-2 Five classifications of NRG cup anemometer with variations in friction
Classification no
Class A
Class B
Class A
Class B
Category
Category
Category
Category
Horizontal Horizontal
Vector
Vector
5) TRTC, FOI
2.4
8.3
2.7
3.0
tilt, with friction
6) TRTC, FOI
1.0
8.3
0.8
2.2
tilt, no friction
7) TRTC, DEWI
2.4
7.7
2.8
4.8
tilt, with friction
8) TRTC, DEWI
1.9
7.7
2.1
4.4
tilt, no friction
9) IFTC
0.6
7.5
0.5
2.6
no friction
The classifications in Table 6-2 show significant variations, especially for the
Class A categories, with or without friction. The deviations were analysed with
a parameter analysis. For a 4m/s, 3% turbulent wind, and for temperatures 0°C,
40°C and 18°C (the calibration temperature), the density was varied 0.91.35kg/m3. The results are shown in Figure 6-10.
Risø-R-1555(EN)
59
0.02
0
-0.02
Deviation (m/s)
-0.04
-0.06
-0.08
-0.1
-0.12
40degC
18degC
0degC
-0.14
-0.16
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
1.4
Air density (kg/m^3)
Figure 6-10 Simulated deviations due to variations in temperature and air
density of NRG cup anemometer at 4m/s and 3% turbulence
The deviations in Figure 6-10 show a high dependence on temperature. Measured friction on the NRG cup anemometer is shown in Figure 6-11. The measured friction at 40°C does deviate from friction at the lower temperatures, and
this seems to explain the significant variations of Table 6-2. In looking at five
friction runs that were actually made in the tests, Figure 6-12, it is seen that
there is a very high spreading in the data. Whether this is caused by the cup
anemometer or spreading due to the test method is at present not investigated.
Running the classification calculations with a maximum temperature of 30°C
with the FOI tilt angular response gives the classification indices in Table 6-3.
The results are seen to lower the Class A horizontal classification index from
2.4 in Table 6-2 down to 1.1.
Friction NRG
2.00E-04
1.80E-04
Friction (kg*m^2/s^2)
1.60E-04
1.40E-04
-10
1
10
30
40
1.20E-04
1.00E-04
8.00E-05
6.00E-05
4.00E-05
2.00E-05
0.00E+00
0
10
20
30
40
50
60
70
Angular speed (rad/s)
Figure 6-11 Measured friction on NRG cup anemometer
60
Risø-R-1555(EN)
Friction NRG
2.00E-04
1.80E-04
Friction (kg*m^2/s^2)
1.60E-04
1.40E-04
40degC (1)
40degC (2)
40degC (3)
40degC (4)
40degC (5)
1.20E-04
1.00E-04
8.00E-05
6.00E-05
4.00E-05
2.00E-05
0.00E+00
0
10
20
30
40
50
60
70
Angular speed (rad/s)
Figure 6-12 Five measurements of friction on NRG cup anemometer at 40°C
Table 6-3 Classification of NRG cup anemometer with variations in friction up
to maximum 30°C
Classification no
Class A
Class B
Class A
Class B
Category
Category
Category
Category
Horizontal Horizontal
Vector
Vector
10) TRTC, FOI
1.1
8.3
1.2
3.0
tilt, with friction
To max 30degC
These classification examples show that the classification index may change
significantly by changes in measurements and modeling. In this respect it
should be noted that there do not exist standards or recommendations of the
measurement procedures of angular response, torque coefficients and friction in
the same way as for normal calibrations. More common use of these measurement procedures, Round Robin exercises between testing institutes, and more
common use of the cup anemometer models may reduce the classification index
variations. On the other hand, the examples demonstrate that the classification
process is very efficient in detecting which characteristics has the most influence on the class index of individual cup anemometer. One general weak characteristic of most cup anemometers is the rather high gradients on angular response around horizontal flow. Small changes in the angular response do immediately influence the Class A index. Some cup anemometer designs also very
efficiently demonstrate insensitivity to friction. The classification tools do in
this way also demonstrate very efficiently their capability to improve on cup
anemometer design.
6.6 Free Field Comparison Tests
6.6.1
Free Field Tests Compared to Simulations
The ultimate verification of the IFTC model capability to predict the cup anemometer behaviour would be to compare detailed field measurements with calculated anemometer response. This requires that the 3D wind field is accurately
Risø-R-1555(EN)
61
known in detail both at the cup anemometer position and at the reference position. Such measurements have not been possible to perform in this project. The
only available simultaneous measurements with cup anemometer and 3D sonic
data have been taken from the comparable measurements of two sonics, ref. 26.
mounted on a boom together with a Risø anemometer. From these measurements the cup/sonic response ratio has been derived by binning the ratio as
function of turbulence intensity.
Neither the details of the spectral content from the boom measurements nor the
turbulence components have been evaluated. Instead it is assumed that the turbulence components correspond to a flat site with ratios 1/0.8/0.5 between the u,
v and w turbulence components. In order to enable a comparison the measured
cup/sonic response ratio also had to be adjusted such that this ratio was approaching unity when going from higher to lower turbulence intensities. The
graph below shows the “comparison”.
Calculated and Meaqsured Anemometer Response in Turbulent Wind (1.0, 0.8, 0.5)
1,01
RIS_x_0.8x_0.5x
Ratio .
Boom measurements Risoe-Cup/Sonic-R3
1,00
0,99
0
5
10
15
20
25
X=Turbulence % (u-comp)
Figure 6-13 “Comparison” of measured and calculated response to turbulence for the Risø anemometer.
6.6.2
Influence of Free Field Horizontal Shear Flow
All models do make assumptions that in the end limit their usefulness. The
models described in the earlier chapters are limited in their description of the
flow field that drives the aerodynamic torque of the cup anemometers. The calibrations, statically as well as dynamically, are assuming uniform flow across the
area of the cup anemometers. The flow in the models is also considered uniform
when an external artificial wind field is applied on the anemometers, and it is
considered a point measurement, representing the centre of the instrument. In
the free field the flow is not uniform. The flow consists of shears and vortices in
all directions, and these influence on the aerodynamic torque of the cup anemometers. A shear that is applied horizontally across the frontal area of a cup
anemometer is specifically influential to the cup anemometer as the shear applies different resulting forces on either side of the cup anemometer. A study of
such shear flow was made by Dahlberg [24] where the shear effect around
masts was studied. The effect was studied in theory applying a drag model of
62
Risø-R-1555(EN)
the cup anemometer. A shear flow with a relative wind speed difference ε at the
centre of the advancing cup and −ε at the centre of the retrogressive cup is applied in the model:
Q A = R( D H − D L ) =
1
ρ AR((U (1 − ε ) − Rω )2C DH − (U (1 + ε ) + Rω )2C DL )
2
Rearranging the terms, it is seen that the relative wind speed difference can be
considered as an addition to the speed ratio:
Q A = R( D H − D L ) =
1
ρ ARU 2 (( 1 − (λ + ε ) )2C DH − ( 1 + (λ + ε ) )2C DL )
2
This means that the speed ratio changes with the relative wind speed difference
ε , and the equilibrium speed ratio changes for a constant wind as:
λ0 =
1− k
−ε
1+ k
The angular speed of the cup anemometer will thus change to:
ω=
U
(λ0 + ε )
R
The change in angular speed corresponds also to a change in an equivalent planar wind speed:
U=
ωR
λ0 + ε
This shear effect leads to small variations in the measurements in a mast
whether a cup anemometer is positioned on the one or the other side of the mast
when the boom is perpendicular to the wind. A positive shear on one side will
give a little higher value, while a negative shear on the other side will give a
little lower value. This effect is opposed when the rotational direction of the cup
anemometer is changed.
An experiment was made in a wind tunnel [25] where an ordinary fine fishing
net with a mesh width of 8.5mm and a thread thickness of 0.15 mm was inserted
in half the test section. A cup anemometer was then traversed behind the net and
the wind speed of the cup anemometer was measured. The measurements
showed a 1,5% lower wind speed when the cup anemometer was fully behind
the net relative to free flow, but when partially behind the net a difference of
about 2.5% to 2.9% was found. For a typical speed ratio λ0 of 0.3 and an ε
value of 0.75%, the difference is calculated to be 2.5%. This corresponds quite
well with the measurements in the wind tunnel.
7. Conclusions
The present report addresses cup anemometer classification methods, based on
the requirements in the IEC61400-12-1 standard on power performance measurements [7]. An introduction of the ideas behind the classification procedure is
outlined in chapter 2, and the requirements for external operational conditions
are examined in detail in chapter 5. Chapter 3 presents a range of different testing methods that all support robust assessment of the cup anemometers. Some
basic testing methods are used to provide data to be fitted to cup anemometer
models, the IFTC and the TRTC models (Chapter 4), and other testing methods
are used for verification of the tuned models to “check” that they are able to
Risø-R-1555(EN)
63
simulate operational conditions realistically. The basic testing methods are:
normal calibration, angular response measurements, static torque coefficient
measurements, and measurement of friction.
An additional new dynamic torque coefficient test method was developed. It
involves exposure of the anemometers to wind gusts in a wind tunnel together
with accurate measurements of the instantaneous wind speed and of the rotational speed of the cup-anemometer. The method requires no attachment of a
rod to the rotor as for the static torque coefficient measurement. This new
method was compared to the more traditional static torque coefficient measurement procedure, and very good agreement was found between the two methods.
The new dynamic torque coefficient test method has the advantage that torque
can be measured under inclined flow by tilting the cup anemometer in the wind
tunnel. This has given rise to the new cup anemometer model, the IFTC model,
which bases the aerodynamic torque on dynamic torque coefficient measurements alone, and do not include angular response measurements in the model.
The robustness of the IFTC model was verified by comparing simulations with
ordinary gust runs in the wind tunnel, and very good agreement was found.
Simulations were also compared with step response measurements with good
result. The inclined flow torque coefficient curves of the IFTC model was used
to compare static angular response measurements. This comparison showed no
deviations between the data. This verifies that angular response data obtained by
static measurements are similar to data obtained by dynamic measurements.
The application of the procedures has been examined in Chapter 6. The robustness of the classification process depends on robustness of the calibration procedures, robustness of the models, and general robustness of application. The
calibration of angular response was compared with measurements of FOI and
DEWI. Significant differences were found in the measured data. The causes for
the deviations in the calibrations have not been examined in detail. FOI is using
a closed test section and the anemometer is moved back and forth in the test
section to obtain different tilted positions. The wind tunnel used by DEWI is an
open test section and the anemometer rotor centre is kept fixed for all tilted positions. The differences influence significantly on the results of classification,
and are the main reason for spreading of classification index data.
The robustness of the models was demonstrated with the IFTC model by verification of simulations to a step response and to response from gust runs. An example of verification of the TRTC model was made for a normal calibration.
The robustness of classification was exemplified by a comparison of four classifications made with both the TRTC and IFTC models on one type of cup anemometer. The results showed variations of Class A index of 1.3-1.9 and of
Class B index of 5.0-8.0. The main reason for these deviations was found in
angular response differences measured at DEWI and FOI.
Another classification example of five simulations on another cup anemometer
type was made with inclusion and exclusion of friction. The result showed
variations of Class A index of 0.6-2.4 and of Class B index of 7.5-8.3. The large
variations in Class A index was found to be due to measured friction variations
at 40°C. Reduction of the temperature variation to 30°C reduced the Class A
index from 2.4 to 1.1.
64
Risø-R-1555(EN)
The two cup anemometer models were used to make classification of five different cup anemometers. The results are presented in a separate report [8].
The results of this classification according to the horizontal wind speed definition were:
o NRG
in Class A 0.6-2.4,
in Class B 7.5-8.3
o Risø
in Class A 1.3-1.9,
in Class B 5.0-8.0
o Thies FC
in Class A 1.5-1.8,
in Class B 2.9-3.8
o Vaisala
in Class A 1.6-2.4,
in Class B 11.0-11.9
o Vector
in Class A 1.3-1.8,
in Class B 4.0-4.5
The classification of the five cup anemometers shows some variation. The
variations have more or less been explained by the exemplifications made for
assessment of robustness. The results are based on single examples of the type
of cup anemometer. The IEC61400-12-1 standard requires at least two examples of a type of cup anemometer to be tested.
Risø-R-1555(EN)
65
8. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
IEC 60044-2:1997, Instrument transformers - Part 2: Inductive voltage
transformers, Amendment 1 (2000)
[12]
Pedersen, T.F. “Development of a Classification System for Cup Anemometers – CLASSCUP”, Risø-R-1348(EN), December 2002
Pedersen, T.F. “On Wind Turbine Power Performance Measurements at
Inclined Airflow”, Wind Energy, 2004; 7:163-176
Pedersen, T.F. “Characterisation and Classification of RISØ P2546 Cup
Anemometer”, Risø-R-1364 (ed.2) (EN), March 2004
IEA Recommended practices for wind turbine testing and evaluation. 3.
Fatigue Loads, edition 1990
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
66
A.Albers, SiteParIden; Relatieve Power Curve Measurements in Flat
Terrain, Final report of Task 6, report JOR3-CT98-0257-DEWI07, 13
June 2001-12-07
A.P.W.M. Curvers and P.A. van der Werff, , SiteParIden; Relative
Power Curve Measurements in Complex Terrain, Final report of Task 7,
report ECN-C—01-102, October 2001.
Albers, A., Klug, H., Westermann, D. ”Outdoor Comparison of Cup
Anemometers”, DEWEK 2000
Albers, A., Klug, H., ”Open Field Cup Anemometry”, EWEC 2000
Dahlberg, J.-Å., Gustavsson, J., Ronsten, G., Pedersen, T.F., Paulsen,
U.S., Westermann, D., “Development of a Standardised Cup Anemometer suited to Wind Energy Applications”, Contract JOR3-CT980263, Publishable Final Report; June 2001
Dahlberg, J.-Å., Pedersen, T.F., Westermann, D., “Development of a
Standardised Cup Anemometer suited to Wind Energy Applications,
CLASSCUP”, EWEC 2000
IEC 61400-12-1, first edition 2005-12, Wind turbines – Part 12-1:
“Power performance measurements of electricity producing wind turbines”
Dahlberg, J-Å, Pedersen, TF, “ACCUWIND – Classification of Five
Cup Anemometers According to IEC61400-12-1”, Risø-R-1556(EN),
May 2006
IEC 60688: 1992, Electrical measuring transducers for converting a.c.
electrical quantities to analogue or digital signals, Amendment 1 (1997),
Amendment 2 (2001)
IEC 60044-1:1996, Instrument transformers - Part 1: Current transformers, Amendment 1 (2000)
Schlichting “Aerodynamic Drag”, page
Schrenck O. “Über die Trägheitsfehler des Schalenkreuz-Anemometers
bei schwankender Windstärke”, Zeitschrift f. techn. Physik, 1929
IEC 61400-1, second edition 1999 “Wind turbine generator systems –
Part 1: Safety requirements”
Frandsen S., Chacón L., Crespo A., Enevoldsen P., Gómez-Elvira R.,
Hernández J., Højstrup J., Manuel F., Thomsen K. “Measurements on
and modelling of offshore wind farms”, Risø-R-903(EN, 1996
ISO 2533 “Standard Atmosphere” 1975
Risø-R-1555(EN)
[21]
[22]
[23]
[24]
[25]
[26]
Mann J “Wind field simulation”, Probabilistic Engineering Mechanics
13(4): 269-282, Oct 1998
Veers, P. S., Three-dimensional wind simulation. Technical Report
SAND88-0152, Sandia National Laboratories, 1988
www.measnet.org
Dahlberg, J-Å, Mast, Boom & Velocity Gradient Effects on Cup Anemometers, FFAP-V-009
Ronsten, G, Dahlberg, J-Å, Calibration of Anemometers in FFA-L2
Wind Tunnel Including Twin Calibration and Tests of Sensitivity to
Wind Velocity Gradient and Tilt, FFA TN 1996-51
Cuerva A, Sanz-Andrés A, Franchini S, Eecen P, Busche P, Pedersen
TF, Mouzakis F, ACCUWIND, Task 2. Improve the Accuracy of Sonic
Anemometers, final report, May 2006
Risø-R-1555(EN)
67
Author: Jan-Åke Dahlberg (FOI), Troels Friis Pedersen (Risø),
Peter Busche (DEWI)
Title: ACCUWIND - Methods for Classification of Cup
Anemometers
Department: Wind Energy Department
Risø-R-1555(EN)
May 2006
Abstract (max. 2000 char.):
ISSN 0106-2840
ISBN 87-550-3514-0
Errors associated with the measurement of wind speed
are the major sources of uncertainties in power
performance testing of wind turbines. Field comparisons
of well-calibrated anemometers show significant and not
acceptable difference. The European CLASSCUP
project posed the objectives to quantify the errors
associated with the use of cup anemometers, and to
develop a classification system for quantification of
systematic errors of cup anemometers. This classification
system has now been implemented in the IEC 61400-121 standard on power performance measurements in
annex I and J. The classification of cup anemometers
requires general external climatic operational ranges to
be applied for the analysis of systematic errors. A Class
A category classification is connected to reasonably flat
sites, and another Class B category is connected to
complex terrain, General classification indices are the
result of assessment of systematic deviations. The
present report focuses on methods that can be applied for
assessment of such systematic deviations. A new
alternative method for torque coefficient measurements
at inclined flow have been developed, which have then
been applied and compared to the existing methods
developed in the CLASSCUP project and earlier. A
number of approaches including the use of two cup
anemometer models, two methods of torque coefficient
measurement, two angular response measurements, and
inclusion and exclusion of influence of friction have
been implemented in the classification process in order
to assess the robustness of methods. The results of the
analysis are presented as classification indices, which are
compared and discussed.
Contract no.:
Group's own reg. no.:
Sponsorship:
Cover :
Pages: 68
Tables: 9
References: 26
Risø National Laboratory
Information Service Department
P.O.Box 49
DK-4000 Roskilde
Denmark
Telephone +45 46774004
[email protected]
Fax +45 46774013
www.risoe.dk
Risø’s research is aimed at solving concrete
problems in the society.
Research targets are set through continuous
dialogue with business, the political system and
researchers.
The effects of our research are sustainable energy
supply and new technology for the health
sector.
www.risoe.dk